[PATCH 0/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[PATCH 0/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[David Rivshin (Allworx)]

From: David Rivshin <drivshin-5fOYsn7Fw8lBDgjK7y7TUQ@public.gmane.org>

This series adds support for the ISSI IS31FL32xx family of I2C LED
controllers. Since the IS31FL3218 is actually the same device as the
SN3218, the dedicated leds-sn3218 driver is removed and the compatible
string is folded into this driver.

Changes from RFC [1]:
 - Removed max-brightness DT property.
 - Added #address-cells and #size-cells properties to the example DT.
 - Refer to these devices as "LED controllers" in Kconfig.
 - Removed redundant last sentence from Kconfig entry
 - Removed unnecessary debug code.
 - Do not set led_classdev.brightness to 0 explicitly, as it is
   already initialized to 0 by devm_kzalloc().
 - Used of_property_read_string() instead of of_get_property().
 - Fail immediately on DT parsing error in a child node, rather than
   continuing on with the non-faulty ones.
 - Added additional comments for some things that might be non-obvious.
 - Added constants for the location of the SSD bit in the SHUTDOWN
   register, and the 3216's CONFIG register.
 - Added special sw_shutdown_func for the 3216 device, as that bit
   is in a different register, at a different position, and has reverse
   polarity compared to all the other devices.
 - Refactored is31fl32xx_init_regs() to separate out some logic into
   is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
 - Added 4th patch to replace the now-redundant leds-sn3218.

[1] http://www.spinics.net/lists/linux-leds/msg05564.html
    http://thread.gmane.org/gmane.linux.leds/4530

David Rivshin (4):
  DT: Add vendor prefix for Integrated Silicon Solutions Inc.
  DT: leds: Add binding for the ISSI IS31FL32xx family of LED
    controllers
  leds: Add driver for the ISSI IS31FL32xx family of LED controllers
  leds: Replace dedicated SN3218 driver with IS31FL32XX driver

 .../devicetree/bindings/leds/leds-is31fl32xx.txt   |  52 +++
 .../devicetree/bindings/leds/leds-sn3218.txt       |  41 --
 .../devicetree/bindings/vendor-prefixes.txt        |   1 +
 drivers/leds/Kconfig                               |  16 +-
 drivers/leds/Makefile                              |   2 +-
 drivers/leds/leds-is31fl32xx.c                     | 509 +++++++++++++++++++++
 drivers/leds/leds-sn3218.c                         | 306 -------------
 7 files changed, 569 insertions(+), 358 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
 delete mode 100644 Documentation/devicetree/bindings/leds/leds-sn3218.txt
 create mode 100644 drivers/leds/leds-is31fl32xx.c
 delete mode 100644 drivers/leds/leds-sn3218.c

-- 
2.5.0

--
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 0/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[David Rivshin (Allworx)]

From: David Rivshin <drivshin@allworx.com>

This series adds support for the ISSI IS31FL32xx family of I2C LED
controllers. Since the IS31FL3218 is actually the same device as the
SN3218, the dedicated leds-sn3218 driver is removed and the compatible
string is folded into this driver.

Changes from RFC [1]:
 - Removed max-brightness DT property.
 - Added #address-cells and #size-cells properties to the example DT.
 - Refer to these devices as "LED controllers" in Kconfig.
 - Removed redundant last sentence from Kconfig entry
 - Removed unnecessary debug code.
 - Do not set led_classdev.brightness to 0 explicitly, as it is
   already initialized to 0 by devm_kzalloc().
 - Used of_property_read_string() instead of of_get_property().
 - Fail immediately on DT parsing error in a child node, rather than
   continuing on with the non-faulty ones.
 - Added additional comments for some things that might be non-obvious.
 - Added constants for the location of the SSD bit in the SHUTDOWN
   register, and the 3216's CONFIG register.
 - Added special sw_shutdown_func for the 3216 device, as that bit
   is in a different register, at a different position, and has reverse
   polarity compared to all the other devices.
 - Refactored is31fl32xx_init_regs() to separate out some logic into
   is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
 - Added 4th patch to replace the now-redundant leds-sn3218.

[1] http://www.spinics.net/lists/linux-leds/msg05564.html
    http://thread.gmane.org/gmane.linux.leds/4530

David Rivshin (4):
  DT: Add vendor prefix for Integrated Silicon Solutions Inc.
  DT: leds: Add binding for the ISSI IS31FL32xx family of LED
    controllers
  leds: Add driver for the ISSI IS31FL32xx family of LED controllers
  leds: Replace dedicated SN3218 driver with IS31FL32XX driver

 .../devicetree/bindings/leds/leds-is31fl32xx.txt   |  52 +++
 .../devicetree/bindings/leds/leds-sn3218.txt       |  41 --
 .../devicetree/bindings/vendor-prefixes.txt        |   1 +
 drivers/leds/Kconfig                               |  16 +-
 drivers/leds/Makefile                              |   2 +-
 drivers/leds/leds-is31fl32xx.c                     | 509 +++++++++++++++++++++
 drivers/leds/leds-sn3218.c                         | 306 -------------
 7 files changed, 569 insertions(+), 358 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
 delete mode 100644 Documentation/devicetree/bindings/leds/leds-sn3218.txt
 create mode 100644 drivers/leds/leds-is31fl32xx.c
 delete mode 100644 drivers/leds/leds-sn3218.c

-- 
2.5.0

[PATCH 1/4] DT: Add vendor prefix for Integrated Silicon Solutions Inc.

[David Rivshin (Allworx)]

From: David Rivshin <drivshin@allworx.com>

ISSI is the stock ticker Integrated Silicon Solutions Inc.
Company website: http://www.issi.com

Signed-off-by: David Rivshin <drivshin@allworx.com>
---

Changes from RFC:
 none

 Documentation/devicetree/bindings/vendor-prefixes.txt | 1 +
 1 file changed, 1 insertion(+)

diff --git a/Documentation/devicetree/bindings/vendor-prefixes.txt b/Documentation/devicetree/bindings/vendor-prefixes.txt
index 52f22f9..dd72e05 100644
--- a/Documentation/devicetree/bindings/vendor-prefixes.txt
+++ b/Documentation/devicetree/bindings/vendor-prefixes.txt
@@ -120,6 +120,7 @@ intercontrol	Inter Control Group
 invensense	InvenSense Inc.
 isee	ISEE 2007 S.L.
 isil	Intersil
+issi	Integrated Silicon Solutions Inc.
 jedec	JEDEC Solid State Technology Association
 karo	Ka-Ro electronics GmbH
 keymile	Keymile GmbH
-- 
2.5.0

[PATCH 2/4] DT: leds: Add binding for the ISSI IS31FL32xx family of LED controllers

[David Rivshin (Allworx)]

From: David Rivshin <drivshin@allworx.com>

This adds a binding description for the is31fl3236/35/18/16 I2C LED
controllers.

Signed-off-by: David Rivshin <drivshin@allworx.com>
---

Rob,
 I went with the 1-based 'reg' property here. I inferred that that
would be your preference based on the previous thread [1]. Let me
know if that's not the case.

Changes from RFC:
 - Removed max-brightness property.
 - Added #address-cells and #size-cells properties to the example.

[1] http://www.spinics.net/lists/linux-leds/msg05589.html

 .../devicetree/bindings/leds/leds-is31fl32xx.txt   | 49 ++++++++++++++++++++++
 1 file changed, 49 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt

diff --git a/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
new file mode 100644
index 0000000..539df2e
--- /dev/null
+++ b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
@@ -0,0 +1,49 @@
+Binding for ISSI IS31FL32xx LED Drivers
+
+The IS31FL32xx family of LED drivers are I2C devices with multiple
+constant-current channels, each with independent 256-level PWM control.
+Each LED is represented as a sub-node of the device.
+
+Required properties:
+- compatible: one of
+	issi,is31fl3236
+	issi,is31fl3235
+	issi,is31fl3218
+	issi,is31fl3216
+- reg: I2C slave address
+- address-cells : must be 1
+- size-cells : must be 0
+
+LED sub-node properties:
+- reg : LED channel number (1..N)
+- label :  (optional)
+  see Documentation/devicetree/bindings/leds/common.txt
+- linux,default-trigger :  (optional)
+  see Documentation/devicetree/bindings/leds/common.txt
+
+
+Example:
+
+leds: is31fl3236@3c {
+	compatible = "issi,is31fl3236";
+	reg = <0x3c>;
+	#address-cells = <1>;
+	#size-cells = <0>;
+
+	led@1 {
+		reg = <1>;
+		label = "EB:blue:usr0";
+	};
+	led@2 {
+		reg = <2>;
+		label = "EB:blue:usr1";
+	};
+	...
+	led@36 {
+		reg = <36>;
+		label = "EB:blue:usr35";
+	};
+};
+
+For more product information please see the link below:
+http://www.issi.com/US/product-analog-fxled-driver.shtml
-- 
2.5.0

[PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[David Rivshin (Allworx)]

From: David Rivshin <drivshin@allworx.com>

The IS31FL32xx family of LED controllers are I2C devices with multiple
constant-current channels, each with independent 256-level PWM control.

Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml

This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
(TI am335x) platform.

The programming paradigm of these devices is similar in the following
ways:
 - All registers are 8 bit
 - All LED control registers are write-only
 - Each LED channel has a PWM register (0-255)
 - PWM register writes are shadowed until an Update register is poked
 - All have a concept of Software Shutdown, which disables output

However, there are some differences in devices:
 - 3236/3235 have a separate Control register for each LED,
   (3218/3216 pack the enable bits into fewer registers)
 - 3236/3235 have a per-channel current divisor setting
 - 3236/3235 have a Global Control register that can turn off all LEDs
 - 3216 is unique in a number of ways
    - OUT9-OUT16 can be configured as GPIOs instead of LED controls
    - LEDs can be programmed with an 8-frame animation, with
      programmable delay between frames
    - LEDs can be modulated by an input audio signal
    - Max output current can be adjusted from 1/4 to 2x globally
    - Has a Configuration register instead of a Shutdown register

This driver currently only supports the base PWM control function
of these devices. The following features of these devices are not
implemented, although it should be possible to add them in the future:
 - All devices are capable of going into a lower-power "software
   shutdown" mode.
 - The is31fl3236 and is31fl3235 can reduce the max output current
   per-channel with a divisor of 1, 2, 3, or 4.
 - The is31fl3216 can use some LED channels as GPIOs instead.
 - The is31fl3216 can animate LEDs in hardware.
 - The is31fl3216 can modulate LEDs according to an audio input.
 - The is31fl3216 can reduce/increase max output current globally.

Signed-off-by: David Rivshin <drivshin@allworx.com>
---

You may see two instances of this warning:
  "passing argument 1 of 'of_property_read_string' discards 'const'
   qualifier from pointer target type"
That is a result of of_property_read_string() taking a non-const
struct device_node pointer parameter. I have separately submitted a
patch to fix that [1], and a few related functions which had the same
issue. I'm hoping that will get into linux-next before this does, so
that the warnings never show up there.

Changes from RFC:
 - Removed max-brightness DT property.
 - Refer to these devices as "LED controllers" in Kconfig.
 - Removed redundant last sentence from Kconfig entry
 - Removed unnecessary debug code.
 - Do not set led_classdev.brightness to 0 explicitly, as it is
   already initialized to 0 by devm_kzalloc().
 - Used of_property_read_string() instead of of_get_property().
 - Fail immediately on DT parsing error in a child node, rather than
   continuing on with the non-faulty ones.
 - Added additional comments for some things that might be non-obvious.
 - Added constants for the location of the SSD bit in the SHUTDOWN
   register, and the 3216's CONFIG register.
 - Added special sw_shutdown_func for the 3216 device, as that bit
   is in a different register, at a different position, and has reverse
   polarity compared to all the other devices.
 - Refactored is31fl32xx_init_regs() to separate out some logic into
   is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().

[1] https://lkml.org/lkml/2016/3/2/746

 drivers/leds/Kconfig           |   8 +
 drivers/leds/Makefile          |   1 +
 drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
 3 files changed, 514 insertions(+)
 create mode 100644 drivers/leds/leds-is31fl32xx.c

diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
index 1034696..9c63ba4 100644
--- a/drivers/leds/Kconfig
+++ b/drivers/leds/Kconfig
@@ -580,6 +580,14 @@ config LEDS_SN3218
 	  This driver can also be built as a module. If so the module
 	  will be called leds-sn3218.
 
+config LEDS_IS31FL32XX
+	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
+	depends on LEDS_CLASS && I2C && OF
+	help
+	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
+	  They are I2C devices with multiple constant-current channels, each
+	  with independent 256-level PWM control.
+
 comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
 
 config LEDS_BLINKM
diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
index 89c9b6f..3fdf313 100644
--- a/drivers/leds/Makefile
+++ b/drivers/leds/Makefile
@@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
 obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
 obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
 obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
+obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
 
 # LED SPI Drivers
 obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
new file mode 100644
index 0000000..49818f0
--- /dev/null
+++ b/drivers/leds/leds-is31fl32xx.c
@@ -0,0 +1,505 @@
+/*
+ * linux/drivers/leds-is31fl32xx.c
+ *
+ * Driver for ISSI IS31FL32xx family of I2C LED controllers
+ *
+ * Copyright 2015 Allworx Corp.
+ *
+ *
+ * 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.
+ *
+ * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
+ */
+
+#include <linux/err.h>
+#include <linux/i2c.h>
+#include <linux/kernel.h>
+#include <linux/leds.h>
+#include <linux/module.h>
+#include <linux/of_platform.h>
+
+/* Used to indicate a device has no such register */
+#define IS31FL32XX_REG_NONE 0xFF
+
+/* Software Shutdown bit in Shutdown Register */
+#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
+#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
+
+/* IS31FL3216 has a number of unique registers */
+#define IS31FL3216_CONFIG_REG 0x00
+#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
+#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
+
+/* Software Shutdown bit in 3216 Config Register */
+#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
+#define IS31FL3216_CONFIG_SSD_DISABLE 0
+
+struct is31fl32xx_priv;
+struct is31fl32xx_led_data {
+	struct led_classdev cdev;
+	u8 channel; /* 1-based, max priv->cdef->channels */
+	struct is31fl32xx_priv *priv;
+};
+
+struct is31fl32xx_priv {
+	const struct is31fl32xx_chipdef *cdef;
+	struct i2c_client *client;
+	unsigned int num_leds;
+	struct is31fl32xx_led_data leds[0];
+};
+
+/**
+ * struct is31fl32xx_chipdef - chip-specific attributes
+ * @channels            : Number of LED channels
+ * @shutdown_reg        : address of Shutdown register (optional)
+ * @pwm_update_reg      : address of PWM Update register
+ * @global_control_reg  : address of Global Control register (optional)
+ * @reset_reg           : address of Reset register (optional)
+ * @pwm_register_base   : address of first PWM register
+ * @pwm_registers_reversed: : true if PWM registers count down instead of up
+ * @led_control_register_base : address of first LED control register (optional)
+ * @enable_bits_per_led_control_register: number of LEDs enable bits in each
+ * @reset_func:         : pointer to reset function
+ *
+ * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
+ * indicates that this chip has no such register.
+ *
+ * If non-NULL, @reset_func will be called during probing to set all
+ * necessary registers to a known initialization state. This is needed
+ * for chips that do not have a @reset_reg.
+ *
+ * @enable_bits_per_led_control_register must be >=1 if
+ * @led_control_register_base != %IS31FL32XX_REG_NONE.
+ */
+struct is31fl32xx_chipdef {
+	u8	channels;
+	u8	shutdown_reg;
+	u8	pwm_update_reg;
+	u8	global_control_reg;
+	u8	reset_reg;
+	u8	pwm_register_base;
+	bool	pwm_registers_reversed;
+	u8	led_control_register_base;
+	u8	enable_bits_per_led_control_register;
+	int (*reset_func)(struct is31fl32xx_priv *priv);
+	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
+};
+
+static const struct is31fl32xx_chipdef is31fl3236_cdef = {
+	.channels				= 36,
+	.shutdown_reg				= 0x00,
+	.pwm_update_reg				= 0x25,
+	.global_control_reg			= 0x4a,
+	.reset_reg				= 0x4f,
+	.pwm_register_base			= 0x01,
+	.led_control_register_base		= 0x26,
+	.enable_bits_per_led_control_register	= 1,
+};
+
+static const struct is31fl32xx_chipdef is31fl3235_cdef = {
+	.channels				= 28,
+	.shutdown_reg				= 0x00,
+	.pwm_update_reg				= 0x25,
+	.global_control_reg			= 0x4a,
+	.reset_reg				= 0x4f,
+	.pwm_register_base			= 0x05,
+	.led_control_register_base		= 0x2a,
+	.enable_bits_per_led_control_register	= 1,
+};
+
+static const struct is31fl32xx_chipdef is31fl3218_cdef = {
+	.channels				= 18,
+	.shutdown_reg				= 0x00,
+	.pwm_update_reg				= 0x16,
+	.global_control_reg			= IS31FL32XX_REG_NONE,
+	.reset_reg				= 0x17,
+	.pwm_register_base			= 0x01,
+	.led_control_register_base		= 0x13,
+	.enable_bits_per_led_control_register	= 6,
+};
+
+static int is31fl3216_reset(struct is31fl32xx_priv *priv);
+static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
+					bool enable);
+static const struct is31fl32xx_chipdef is31fl3216_cdef = {
+	.channels				= 16,
+	.shutdown_reg				= IS31FL32XX_REG_NONE,
+	.pwm_update_reg				= 0xB0,
+	.global_control_reg			= IS31FL32XX_REG_NONE,
+	.reset_reg				= IS31FL32XX_REG_NONE,
+	.pwm_register_base			= 0x10,
+	.pwm_registers_reversed			= true,
+	.led_control_register_base		= 0x01,
+	.enable_bits_per_led_control_register	= 8,
+	.reset_func				= is31fl3216_reset,
+	.sw_shutdown_func			= is31fl3216_software_shutdown,
+};
+
+static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
+{
+	int ret;
+
+	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
+
+	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
+	if (ret) {
+		dev_err(&priv->client->dev,
+			"register write to 0x%02X failed (error %d)",
+			reg, ret);
+	}
+	return ret;
+}
+
+/*
+ * Custom reset function for IS31FL3216 because it does not have a RESET
+ * register the way that the other IS31FL32xx chips do. We don't bother
+ * writing the GPIO and animation registers, because the registers we
+ * do write ensure those will have no effect.
+ */
+static int is31fl3216_reset(struct is31fl32xx_priv *priv)
+{
+	unsigned int i;
+	int ret;
+
+	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
+			       IS31FL3216_CONFIG_SSD_ENABLE);
+	if (ret)
+		return ret;
+	for (i = 0; i < priv->cdef->channels; i++) {
+		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
+				       0x00);
+		if (ret)
+			return ret;
+	}
+	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
+	if (ret)
+		return ret;
+	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
+	if (ret)
+		return ret;
+	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+/*
+ * Custom Software-Shutdown function for IS31FL3216 because it does not have
+ * a SHUTDOWN register the way that the other IS31FL32xx chips do.
+ * We don't bother doing a read/modify/write on the CONFIG register because
+ * we only ever use a value of '0' for the other fields in that register.
+ */
+static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
+					bool enable)
+{
+	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
+			    IS31FL3216_CONFIG_SSD_DISABLE;
+
+	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
+}
+
+/*
+ * NOTE: A mutex is not needed in this function because:
+ * - All referenced data is read-only after probe()
+ * - The I2C core has a mutex on to protect the bus
+ * - There are no read/modify/write operations
+ * - Intervening operations between the write of the PWM register
+ *   and the Update register are harmless.
+ *
+ * Example:
+ *	PWM_REG_1 write 16
+ *	UPDATE_REG write 0
+ *	PWM_REG_2 write 128
+ *	UPDATE_REG write 0
+ *   vs:
+ *	PWM_REG_1 write 16
+ *	PWM_REG_2 write 128
+ *	UPDATE_REG write 0
+ *	UPDATE_REG write 0
+ * are equivalent. Poking the Update register merely applies all PWM
+ * register writes up to that point.
+ */
+static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
+				     enum led_brightness brightness)
+{
+	const struct is31fl32xx_led_data *led_data =
+		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
+	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
+	u8 pwm_register_offset;
+	int ret;
+
+	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
+
+	/* NOTE: led_data->channel is 1-based */
+	if (cdef->pwm_registers_reversed)
+		pwm_register_offset = cdef->channels - led_data->channel;
+	else
+		pwm_register_offset = led_data->channel - 1;
+
+	ret = is31fl32xx_write(led_data->priv,
+			       cdef->pwm_register_base + pwm_register_offset,
+			       brightness);
+	if (ret)
+		return ret;
+
+	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
+}
+
+static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
+{
+	const struct is31fl32xx_chipdef *cdef = priv->cdef;
+	int ret;
+
+	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
+		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
+		if (ret)
+			return ret;
+	}
+
+	if (cdef->reset_func)
+		return cdef->reset_func(priv);
+
+	return 0;
+}
+
+static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
+					bool enable)
+{
+	const struct is31fl32xx_chipdef *cdef = priv->cdef;
+	int ret;
+
+	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
+		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
+				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
+		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
+		if (ret)
+			return ret;
+	}
+
+	if (cdef->sw_shutdown_func)
+		return cdef->sw_shutdown_func(priv, enable);
+
+	return 0;
+}
+
+static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
+{
+	const struct is31fl32xx_chipdef *cdef = priv->cdef;
+	int ret;
+
+	ret = is31fl32xx_reset_regs(priv);
+	if (ret)
+		return ret;
+
+	/*
+	 * Set enable bit for all channels.
+	 * We will control state with PWM registers alone.
+	 */
+	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
+		u8 value =
+		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
+		u8 num_regs = cdef->channels /
+				cdef->enable_bits_per_led_control_register;
+		int i;
+
+		for (i = 0; i < num_regs; i++) {
+			ret = is31fl32xx_write(priv,
+					       cdef->led_control_register_base+i,
+					       value);
+			if (ret)
+				return ret;
+		}
+	}
+
+	ret = is31fl32xx_software_shutdown(priv, false);
+	if (ret)
+		return ret;
+
+	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
+		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static inline size_t sizeof_is31fl32xx_priv(int num_leds)
+{
+	return sizeof(struct is31fl32xx_priv) +
+		      (sizeof(struct is31fl32xx_led_data) * num_leds);
+}
+
+static int is31fl32xx_parse_child_dt(const struct device *dev,
+				     const struct device_node *child,
+				     struct is31fl32xx_led_data *led_data)
+{
+	struct led_classdev *cdev = &led_data->cdev;
+	int ret = 0;
+	u32 reg;
+
+	if (of_property_read_string(child, "label", &cdev->name))
+		cdev->name = child->name;
+
+	ret = of_property_read_u32(child, "reg", &reg);
+	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
+		dev_err(dev,
+			"Child node %s does not have a valid reg property\n",
+			child->full_name);
+		return -EINVAL;
+	}
+	led_data->channel = reg;
+
+	of_property_read_string(child, "linux,default-trigger",
+				&cdev->default_trigger);
+
+	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
+
+	return 0;
+}
+
+static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
+					struct is31fl32xx_priv *priv,
+					u8 channel)
+{
+	size_t i;
+
+	for (i = 0; i < priv->num_leds; i++) {
+		if (priv->leds[i].channel == channel)
+			return &priv->leds[i];
+	}
+
+	return NULL;
+}
+
+static int is31fl32xx_parse_dt(struct device *dev,
+			       struct is31fl32xx_priv *priv)
+{
+	struct device_node *child;
+	int ret = 0;
+
+	for_each_child_of_node(dev->of_node, child) {
+		struct is31fl32xx_led_data *led_data =
+			&priv->leds[priv->num_leds];
+		const struct is31fl32xx_led_data *other_led_data;
+
+		led_data->priv = priv;
+
+		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
+		if (ret)
+			goto err;
+
+		/* Detect if channel is already in use by another child */
+		other_led_data = is31fl32xx_find_led_data(priv,
+							  led_data->channel);
+		if (other_led_data) {
+			dev_err(dev,
+				"%s and %s both attempting to use channel %d\n",
+				led_data->cdev.name,
+				other_led_data->cdev.name,
+				led_data->channel);
+			goto err;
+		}
+
+		ret = devm_led_classdev_register(dev, &led_data->cdev);
+		if (ret) {
+			dev_err(dev, "failed to register PWM led for %s: %d\n",
+				led_data->cdev.name, ret);
+			goto err;
+		}
+
+		priv->num_leds++;
+	}
+
+	return 0;
+
+err:
+	of_node_put(child);
+	return ret;
+}
+
+static const struct of_device_id of_is31fl31xx_match[] = {
+	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
+	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
+	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
+	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
+	{},
+};
+
+MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
+
+static int is31fl32xx_probe(struct i2c_client *client,
+			    const struct i2c_device_id *id)
+{
+	const struct is31fl32xx_chipdef *cdef;
+	const struct of_device_id *of_dev_id;
+	struct device *dev = &client->dev;
+	struct is31fl32xx_priv *priv;
+	int count;
+	int ret = 0;
+
+	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
+	if (!of_dev_id)
+		return -EINVAL;
+
+	cdef = of_dev_id->data;
+
+	count = of_get_child_count(dev->of_node);
+	if (!count)
+		return -EINVAL;
+
+	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
+			    GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	priv->client = client;
+	priv->cdef = cdef;
+	i2c_set_clientdata(client, priv);
+
+	ret = is31fl32xx_init_regs(priv);
+	if (ret)
+		return ret;
+
+	ret = is31fl32xx_parse_dt(dev, priv);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int is31fl32xx_remove(struct i2c_client *client)
+{
+	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
+
+	return is31fl32xx_reset_regs(priv);
+}
+
+/*
+ * i2c-core requires that id_table be non-NULL, even though
+ * it is not used for DeviceTree based instantiation.
+ */
+static const struct i2c_device_id is31fl31xx_id[] = {
+	{},
+};
+
+MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
+
+static struct i2c_driver is31fl32xx_driver = {
+	.driver = {
+		.name	= "is31fl32xx",
+		.of_match_table = of_is31fl31xx_match,
+	},
+	.probe		= is31fl32xx_probe,
+	.remove		= is31fl32xx_remove,
+	.id_table	= is31fl31xx_id,
+};
+
+module_i2c_driver(is31fl32xx_driver);
+
+MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
+MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
+MODULE_LICENSE("GPL v2");
-- 
2.5.0

[PATCH 4/4] leds: Replace dedicated SN3218 driver with IS31FL32XX driver

[David Rivshin (Allworx)]

From: David Rivshin <drivshin@allworx.com>

Si-En Technology was acquired by ISSI in 2011, and it appears that
the IS31FL3218/IS31FL3216 are just rebranded SN3218/SN3216 devices.
As the IS31FL32XX driver already handles the *3218 devices, there
is no longer a need for the dedicated SN3218 driver.

Add the "sn,sn3218" and "sn,sn3216" compatible strings into the
IS31FL32XX driver and binding documentation, and remove the
leds-sn3218 driver.

Datasheets:
    IS31FL3218: http://www.issi.com/WW/pdf/31FL3218.pdf
    SN3218:     http://www.si-en.com/uploadpdf/s2011517171720.pdf

    IS31FL3216: http://www.issi.com/WW/pdf/31FL3216.pdf
    SN3216;     http://www.si-en.com/uploadpdf/SN3216201152410148.pdf

Signed-off-by: David Rivshin <drivshin@allworx.com>
---

Note that the leds-sn3218 binding use a 0-based 'reg' property, while
the leds-is31fl32xx binding uses a 1-based 'reg' property. This seemed
to be the preferred binding based on [1]. Since leds-sn3216 has not been
in a released kernel, there is are no backwards-compatibility concerns.

Changes from RFC:
 new

[1] http://www.spinics.net/lists/linux-leds/msg05589.html

 .../devicetree/bindings/leds/leds-is31fl32xx.txt   |   9 +-
 .../devicetree/bindings/leds/leds-sn3218.txt       |  41 ---
 drivers/leds/Kconfig                               |  18 +-
 drivers/leds/Makefile                              |   1 -
 drivers/leds/leds-is31fl32xx.c                     |   6 +-
 drivers/leds/leds-sn3218.c                         | 306 ---------------------
 6 files changed, 14 insertions(+), 367 deletions(-)
 delete mode 100644 Documentation/devicetree/bindings/leds/leds-sn3218.txt
 delete mode 100644 drivers/leds/leds-sn3218.c

diff --git a/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
index 539df2e..c59eb1a 100644
--- a/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
+++ b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
@@ -1,6 +1,6 @@
-Binding for ISSI IS31FL32xx LED Drivers
+Binding for ISSI IS31FL32xx and Si-En SN32xx LED Drivers
 
-The IS31FL32xx family of LED drivers are I2C devices with multiple
+The IS31FL32xx/SN32xx family of LED drivers are I2C devices with multiple
 constant-current channels, each with independent 256-level PWM control.
 Each LED is represented as a sub-node of the device.
 
@@ -10,6 +10,8 @@ Required properties:
 	issi,is31fl3235
 	issi,is31fl3218
 	issi,is31fl3216
+	si-en,sn3218
+	si-en,sn3216
 - reg: I2C slave address
 - address-cells : must be 1
 - size-cells : must be 0
@@ -45,5 +47,6 @@ leds: is31fl3236@3c {
 	};
 };
 
-For more product information please see the link below:
+For more product information please see the links below:
 http://www.issi.com/US/product-analog-fxled-driver.shtml
+http://www.si-en.com/product.asp?parentid=890
diff --git a/Documentation/devicetree/bindings/leds/leds-sn3218.txt b/Documentation/devicetree/bindings/leds/leds-sn3218.txt
deleted file mode 100644
index 19cbf57..0000000
--- a/Documentation/devicetree/bindings/leds/leds-sn3218.txt
+++ /dev/null
@@ -1,41 +0,0 @@
-* Si-En Technology - SN3218 18-Channel LED Driver
-
-Required properties:
-- compatible :
-	"si-en,sn3218"
-- address-cells : must be 1
-- size-cells : must be 0
-- reg : I2C slave address, typically 0x54
-
-There must be at least 1 LED which is represented as a sub-node
-of the sn3218 device.
-
-LED sub-node properties:
-- label : (optional) see Documentation/devicetree/bindings/leds/common.txt
-- reg : number of LED line (could be from 0 to 17)
-- linux,default-trigger : (optional)
-   see Documentation/devicetree/bindings/leds/common.txt
-
-Example:
-
-sn3218: led-controller@54 {
-	compatible = "si-en,sn3218";
-	#address-cells = <1>;
-	#size-cells = <0>;
-	reg = <0x54>;
-
-	led@0 {
-		label = "led1";
-		reg = <0x0>;
-	};
-
-	led@1 {
-		label = "led2";
-		reg = <0x1>;
-	};
-
-	led@2 {
-		label = "led3";
-		reg = <0x2>;
-	};
-};
diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
index 9c63ba4..1f64151 100644
--- a/drivers/leds/Kconfig
+++ b/drivers/leds/Kconfig
@@ -568,25 +568,13 @@ config LEDS_SEAD3
 	  This driver can also be built as a module. If so the module
 	  will be called leds-sead3.
 
-config LEDS_SN3218
-	tristate "LED support for Si-En SN3218 I2C chip"
-	depends on LEDS_CLASS && I2C
-	depends on OF
-	select REGMAP_I2C
-	help
-	  This option enables support for the Si-EN SN3218 LED driver
-	  connected through I2C. Say Y to enable support for the SN3218 LED.
-
-	  This driver can also be built as a module. If so the module
-	  will be called leds-sn3218.
-
 config LEDS_IS31FL32XX
 	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
 	depends on LEDS_CLASS && I2C && OF
 	help
-	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
-	  They are I2C devices with multiple constant-current channels, each
-	  with independent 256-level PWM control.
+	  Say Y here to include support for ISSI IS31FL32XX and Si-En SN32xx
+	  LED controllers. They are I2C devices with multiple constant-current
+	  channels, each with independent 256-level PWM control.
 
 comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
 
diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
index 3fdf313..cb2013d 100644
--- a/drivers/leds/Makefile
+++ b/drivers/leds/Makefile
@@ -66,7 +66,6 @@ obj-$(CONFIG_LEDS_MENF21BMC)		+= leds-menf21bmc.o
 obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
 obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
 obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
-obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
 obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
 
 # LED SPI Drivers
diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
index 49818f0..ec3f541 100644
--- a/drivers/leds/leds-is31fl32xx.c
+++ b/drivers/leds/leds-is31fl32xx.c
@@ -10,7 +10,9 @@
  * it under the terms of the GNU General Public License version 2 as
  * published by the Free Software Foundation.
  *
- * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
+ * Datasheets:
+ *   http://www.issi.com/US/product-analog-fxled-driver.shtml
+ *   http://www.si-en.com/product.asp?parentid=890
  */
 
 #include <linux/err.h>
@@ -425,7 +427,9 @@ static const struct of_device_id of_is31fl31xx_match[] = {
 	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
 	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
 	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
+	{ .compatible = "si-en,sn3218",    .data = &is31fl3218_cdef, },
 	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
+	{ .compatible = "si-en,sn3216",    .data = &is31fl3216_cdef, },
 	{},
 };
 
diff --git a/drivers/leds/leds-sn3218.c b/drivers/leds/leds-sn3218.c
deleted file mode 100644
index dcc2581..0000000
--- a/drivers/leds/leds-sn3218.c
+++ /dev/null
@@ -1,306 +0,0 @@
-/*
- * Si-En SN3218 18 Channel LED Driver
- *
- * Copyright (C) 2016 Stefan Wahren <stefan.wahren@i2se.com>
- *
- * Based on leds-pca963x.c
- *
- * 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.
- *
- * Datasheet: http://www.si-en.com/uploadpdf/s2011517171720.pdf
- *
- */
-
-#include <linux/err.h>
-#include <linux/i2c.h>
-#include <linux/leds.h>
-#include <linux/module.h>
-#include <linux/of.h>
-#include <linux/regmap.h>
-#include <linux/slab.h>
-
-#define SN3218_MODE		0x00
-#define SN3218_PWM_1		0x01
-#define SN3218_PWM_2		0x02
-#define SN3218_PWM_3		0x03
-#define SN3218_PWM_4		0x04
-#define SN3218_PWM_5		0x05
-#define SN3218_PWM_6		0x06
-#define SN3218_PWM_7		0x07
-#define SN3218_PWM_8		0x08
-#define SN3218_PWM_9		0x09
-#define SN3218_PWM_10		0x0a
-#define SN3218_PWM_11		0x0b
-#define SN3218_PWM_12		0x0c
-#define SN3218_PWM_13		0x0d
-#define SN3218_PWM_14		0x0e
-#define SN3218_PWM_15		0x0f
-#define SN3218_PWM_16		0x10
-#define SN3218_PWM_17		0x11
-#define SN3218_PWM_18		0x12
-#define SN3218_LED_1_6		0x13
-#define SN3218_LED_7_12		0x14
-#define SN3218_LED_13_18	0x15
-#define SN3218_UPDATE		0x16	/* applies to reg 0x01 .. 0x15 */
-#define SN3218_RESET		0x17
-
-#define SN3218_LED_MASK		0x3f
-#define SN3218_LED_ON		0x01
-#define SN3218_LED_OFF		0x00
-
-#define SN3218_MODE_SHUTDOWN	0x00
-#define SN3218_MODE_NORMAL	0x01
-
-#define NUM_LEDS		18
-
-struct sn3218_led;
-
-/**
- * struct sn3218 -
- * @client - Pointer to the I2C client
- * @leds - Pointer to the individual LEDs
- * @num_leds - Actual number of LEDs
-**/
-struct sn3218 {
-	struct i2c_client *client;
-	struct regmap *regmap;
-	struct sn3218_led *leds;
-	int num_leds;
-};
-
-/**
- * struct sn3218_led -
- * @chip - Pointer to the container
- * @led_cdev - led class device pointer
- * @led_num - LED index ( 0 .. 17 )
-**/
-struct sn3218_led {
-	struct sn3218 *chip;
-	struct led_classdev led_cdev;
-	int led_num;
-};
-
-static int sn3218_led_set(struct led_classdev *led_cdev,
-			  enum led_brightness brightness)
-{
-	struct sn3218_led *led =
-			container_of(led_cdev, struct sn3218_led, led_cdev);
-	struct regmap *regmap = led->chip->regmap;
-	u8 bank = led->led_num / 6;
-	u8 mask = 0x1 << (led->led_num % 6);
-	u8 val;
-	int ret;
-
-	if (brightness == LED_OFF)
-		val = 0;
-	else
-		val = mask;
-
-	ret = regmap_update_bits(regmap, SN3218_LED_1_6 + bank, mask, val);
-	if (ret < 0)
-		return ret;
-
-	if (brightness > LED_OFF) {
-		ret = regmap_write(regmap, SN3218_PWM_1 + led->led_num,
-				   brightness);
-		if (ret < 0)
-			return ret;
-	}
-
-	ret = regmap_write(regmap, SN3218_UPDATE, 0xff);
-
-	return ret;
-}
-
-static void sn3218_led_init(struct sn3218 *sn3218, struct device_node *node,
-			    u32 reg)
-{
-	struct sn3218_led *leds = sn3218->leds;
-	struct led_classdev *cdev = &leds[reg].led_cdev;
-
-	leds[reg].led_num = reg;
-	leds[reg].chip = sn3218;
-
-	if (of_property_read_string(node, "label", &cdev->name))
-		cdev->name = node->name;
-
-	of_property_read_string(node, "linux,default-trigger",
-				&cdev->default_trigger);
-
-	cdev->brightness_set_blocking = sn3218_led_set;
-}
-
-static const struct reg_default sn3218_reg_defs[] = {
-	{ SN3218_MODE, 0x00},
-	{ SN3218_PWM_1, 0x00},
-	{ SN3218_PWM_2, 0x00},
-	{ SN3218_PWM_3, 0x00},
-	{ SN3218_PWM_4, 0x00},
-	{ SN3218_PWM_5, 0x00},
-	{ SN3218_PWM_6, 0x00},
-	{ SN3218_PWM_7, 0x00},
-	{ SN3218_PWM_8, 0x00},
-	{ SN3218_PWM_9, 0x00},
-	{ SN3218_PWM_10, 0x00},
-	{ SN3218_PWM_11, 0x00},
-	{ SN3218_PWM_12, 0x00},
-	{ SN3218_PWM_13, 0x00},
-	{ SN3218_PWM_14, 0x00},
-	{ SN3218_PWM_15, 0x00},
-	{ SN3218_PWM_16, 0x00},
-	{ SN3218_PWM_17, 0x00},
-	{ SN3218_PWM_18, 0x00},
-	{ SN3218_LED_1_6, 0x00},
-	{ SN3218_LED_7_12, 0x00},
-	{ SN3218_LED_13_18, 0x00},
-	{ SN3218_UPDATE, 0x00},
-	{ SN3218_RESET, 0x00},
-};
-
-static const struct regmap_config sn3218_regmap_config = {
-	.reg_bits = 8,
-	.val_bits = 8,
-
-	.max_register = SN3218_RESET,
-	.reg_defaults = sn3218_reg_defs,
-	.num_reg_defaults = ARRAY_SIZE(sn3218_reg_defs),
-	.cache_type = REGCACHE_RBTREE,
-};
-
-static int sn3218_init(struct i2c_client *client, struct sn3218 *sn3218)
-{
-	struct device_node *np = client->dev.of_node, *child;
-	struct sn3218_led *leds;
-	int ret, count;
-
-	count = of_get_child_count(np);
-	if (!count)
-		return -ENODEV;
-
-	if (count > NUM_LEDS) {
-		dev_err(&client->dev, "Invalid LED count %d\n", count);
-		return -EINVAL;
-	}
-
-	leds = devm_kzalloc(&client->dev, count * sizeof(*leds), GFP_KERNEL);
-	if (!leds)
-		return -ENOMEM;
-
-	sn3218->leds = leds;
-	sn3218->num_leds = count;
-	sn3218->client = client;
-
-	sn3218->regmap = devm_regmap_init_i2c(client, &sn3218_regmap_config);
-	if (IS_ERR(sn3218->regmap)) {
-		ret = PTR_ERR(sn3218->regmap);
-		dev_err(&client->dev, "Failed to allocate register map: %d\n",
-			ret);
-		return ret;
-	}
-
-	for_each_child_of_node(np, child) {
-		u32 reg;
-
-		ret = of_property_read_u32(child, "reg", &reg);
-		if (ret)
-			goto fail;
-
-		if (reg >= count) {
-			dev_err(&client->dev, "Invalid LED (%u >= %d)\n", reg,
-				count);
-			ret = -EINVAL;
-			goto fail;
-		}
-
-		sn3218_led_init(sn3218, child, reg);
-	}
-
-	return 0;
-
-fail:
-	of_node_put(child);
-	return ret;
-}
-
-static int sn3218_probe(struct i2c_client *client,
-			const struct i2c_device_id *id)
-{
-	struct sn3218 *sn3218;
-	struct sn3218_led *leds;
-	struct device *dev = &client->dev;
-	int i, ret;
-
-	sn3218 = devm_kzalloc(dev, sizeof(*sn3218), GFP_KERNEL);
-	if (!sn3218)
-		return -ENOMEM;
-
-	ret = sn3218_init(client, sn3218);
-	if (ret)
-		return ret;
-
-	i2c_set_clientdata(client, sn3218);
-	leds = sn3218->leds;
-
-	/*
-	 * Since the chip is write-only we need to reset him into
-	 * a defined state (all LEDs off).
-	 */
-	ret = regmap_write(sn3218->regmap, SN3218_RESET, 0xff);
-	if (ret)
-		return ret;
-
-	for (i = 0; i < sn3218->num_leds; i++) {
-		ret = devm_led_classdev_register(dev, &leds[i].led_cdev);
-		if (ret < 0)
-			return ret;
-	}
-
-	return regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_NORMAL);
-}
-
-static int sn3218_remove(struct i2c_client *client)
-{
-	struct sn3218 *sn3218 = i2c_get_clientdata(client);
-
-	regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_SHUTDOWN);
-
-	return 0;
-}
-
-static void sn3218_shutdown(struct i2c_client *client)
-{
-	struct sn3218 *sn3218 = i2c_get_clientdata(client);
-
-	regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_SHUTDOWN);
-}
-
-static const struct i2c_device_id sn3218_id[] = {
-	{ "sn3218", 0 },
-	{ }
-};
-MODULE_DEVICE_TABLE(i2c, sn3218_id);
-
-static const struct of_device_id of_sn3218_match[] = {
-	{ .compatible = "si-en,sn3218", },
-	{},
-};
-MODULE_DEVICE_TABLE(of, of_sn3218_match);
-
-static struct i2c_driver sn3218_driver = {
-	.driver = {
-		.name	= "leds-sn3218",
-		.of_match_table = of_match_ptr(of_sn3218_match),
-	},
-	.probe	= sn3218_probe,
-	.remove	= sn3218_remove,
-	.shutdown = sn3218_shutdown,
-	.id_table = sn3218_id,
-};
-
-module_i2c_driver(sn3218_driver);
-
-MODULE_DESCRIPTION("Si-En SN3218 LED Driver");
-MODULE_AUTHOR("Stefan Wahren <stefan.wahren@i2se.com>");
-MODULE_LICENSE("GPL v2");
-- 
2.5.0

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[kbuild test robot]

[-- Attachment #1: Type: text/plain, Size: 3397 bytes --]

Hi David,

[auto build test WARNING on j.anaszewski-leds/for-next]
[also build test WARNING on next-20160302]
[cannot apply to v4.5-rc6]
[if your patch is applied to the wrong git tree, please drop us a note to help improving the system]

url:    https://github.com/0day-ci/linux/commits/David-Rivshin-Allworx/leds-Add-driver-for-the-ISSI-IS31FL32xx-family-of-LED-controllers/20160303-110656
base:   https://git.kernel.org/pub/scm/linux/kernel/git/j.anaszewski/linux-leds for-next
config: sparc64-allyesconfig (attached as .config)
reproduce:
        wget https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # save the attached .config to linux build tree
        make.cross ARCH=sparc64 

All warnings (new ones prefixed by >>):

   drivers/leds/leds-is31fl32xx.c: In function 'is31fl32xx_parse_child_dt':
>> drivers/leds/leds-is31fl32xx.c:344:6: warning: passing argument 1 of 'of_property_read_string' discards 'const' qualifier from pointer target type
     if (of_property_read_string(child, "label", &cdev->name))
         ^
   In file included from arch/sparc/include/asm/openprom.h:14:0,
                    from arch/sparc/include/asm/device.h:9,
                    from include/linux/device.h:30,
                    from include/linux/i2c.h:30,
                    from drivers/leds/leds-is31fl32xx.c:17:
   include/linux/of.h:299:12: note: expected 'struct device_node *' but argument is of type 'const struct device_node *'
    extern int of_property_read_string(struct device_node *np,
               ^
   drivers/leds/leds-is31fl32xx.c:356:2: warning: passing argument 1 of 'of_property_read_string' discards 'const' qualifier from pointer target type
     of_property_read_string(child, "linux,default-trigger",
     ^
   In file included from arch/sparc/include/asm/openprom.h:14:0,
                    from arch/sparc/include/asm/device.h:9,
                    from include/linux/device.h:30,
                    from include/linux/i2c.h:30,
                    from drivers/leds/leds-is31fl32xx.c:17:
   include/linux/of.h:299:12: note: expected 'struct device_node *' but argument is of type 'const struct device_node *'
    extern int of_property_read_string(struct device_node *np,
               ^

vim +344 drivers/leds/leds-is31fl32xx.c

   328	}
   329	
   330	static inline size_t sizeof_is31fl32xx_priv(int num_leds)
   331	{
   332		return sizeof(struct is31fl32xx_priv) +
   333			      (sizeof(struct is31fl32xx_led_data) * num_leds);
   334	}
   335	
   336	static int is31fl32xx_parse_child_dt(const struct device *dev,
   337					     const struct device_node *child,
   338					     struct is31fl32xx_led_data *led_data)
   339	{
   340		struct led_classdev *cdev = &led_data->cdev;
   341		int ret = 0;
   342		u32 reg;
   343	
 > 344		if (of_property_read_string(child, "label", &cdev->name))
   345			cdev->name = child->name;
   346	
   347		ret = of_property_read_u32(child, "reg", &reg);
   348		if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
   349			dev_err(dev,
   350				"Child node %s does not have a valid reg property\n",
   351				child->full_name);
   352			return -EINVAL;

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation

[-- Attachment #2: .config.gz --]
[-- Type: application/octet-stream, Size: 45112 bytes --]

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[Jacek Anaszewski]

Hi David,

Thanks for the update. Two remarks in the code.

On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:
> From: David Rivshin <drivshin@allworx.com>
>
> The IS31FL32xx family of LED controllers are I2C devices with multiple
> constant-current channels, each with independent 256-level PWM control.
>
> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>
> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
> (TI am335x) platform.
>
> The programming paradigm of these devices is similar in the following
> ways:
>   - All registers are 8 bit
>   - All LED control registers are write-only
>   - Each LED channel has a PWM register (0-255)
>   - PWM register writes are shadowed until an Update register is poked
>   - All have a concept of Software Shutdown, which disables output
>
> However, there are some differences in devices:
>   - 3236/3235 have a separate Control register for each LED,
>     (3218/3216 pack the enable bits into fewer registers)
>   - 3236/3235 have a per-channel current divisor setting
>   - 3236/3235 have a Global Control register that can turn off all LEDs
>   - 3216 is unique in a number of ways
>      - OUT9-OUT16 can be configured as GPIOs instead of LED controls
>      - LEDs can be programmed with an 8-frame animation, with
>        programmable delay between frames
>      - LEDs can be modulated by an input audio signal
>      - Max output current can be adjusted from 1/4 to 2x globally
>      - Has a Configuration register instead of a Shutdown register
>
> This driver currently only supports the base PWM control function
> of these devices. The following features of these devices are not
> implemented, although it should be possible to add them in the future:
>   - All devices are capable of going into a lower-power "software
>     shutdown" mode.
>   - The is31fl3236 and is31fl3235 can reduce the max output current
>     per-channel with a divisor of 1, 2, 3, or 4.
>   - The is31fl3216 can use some LED channels as GPIOs instead.
>   - The is31fl3216 can animate LEDs in hardware.
>   - The is31fl3216 can modulate LEDs according to an audio input.
>   - The is31fl3216 can reduce/increase max output current globally.
>
> Signed-off-by: David Rivshin <drivshin@allworx.com>
> ---
>
> You may see two instances of this warning:
>    "passing argument 1 of 'of_property_read_string' discards 'const'
>     qualifier from pointer target type"
> That is a result of of_property_read_string() taking a non-const
> struct device_node pointer parameter. I have separately submitted a
> patch to fix that [1], and a few related functions which had the same
> issue. I'm hoping that will get into linux-next before this does, so
> that the warnings never show up there.

Please adjust the patch so that it compiles without warnings on
current linux-next. Your patch for DT API hasn't been reviewed yet
AFICS, and I can imagine that there will be some resistance against.

> Changes from RFC:
>   - Removed max-brightness DT property.
>   - Refer to these devices as "LED controllers" in Kconfig.
>   - Removed redundant last sentence from Kconfig entry
>   - Removed unnecessary debug code.
>   - Do not set led_classdev.brightness to 0 explicitly, as it is
>     already initialized to 0 by devm_kzalloc().
>   - Used of_property_read_string() instead of of_get_property().
>   - Fail immediately on DT parsing error in a child node, rather than
>     continuing on with the non-faulty ones.
>   - Added additional comments for some things that might be non-obvious.
>   - Added constants for the location of the SSD bit in the SHUTDOWN
>     register, and the 3216's CONFIG register.
>   - Added special sw_shutdown_func for the 3216 device, as that bit
>     is in a different register, at a different position, and has reverse
>     polarity compared to all the other devices.
>   - Refactored is31fl32xx_init_regs() to separate out some logic into
>     is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
>
> [1] https://lkml.org/lkml/2016/3/2/746
>
>   drivers/leds/Kconfig           |   8 +
>   drivers/leds/Makefile          |   1 +
>   drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
>   3 files changed, 514 insertions(+)
>   create mode 100644 drivers/leds/leds-is31fl32xx.c
>
> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> index 1034696..9c63ba4 100644
> --- a/drivers/leds/Kconfig
> +++ b/drivers/leds/Kconfig
> @@ -580,6 +580,14 @@ config LEDS_SN3218
>   	  This driver can also be built as a module. If so the module
>   	  will be called leds-sn3218.
>
> +config LEDS_IS31FL32XX
> +	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
> +	depends on LEDS_CLASS && I2C && OF
> +	help
> +	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
> +	  They are I2C devices with multiple constant-current channels, each
> +	  with independent 256-level PWM control.
> +
>   comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
>
>   config LEDS_BLINKM
> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
> index 89c9b6f..3fdf313 100644
> --- a/drivers/leds/Makefile
> +++ b/drivers/leds/Makefile
> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
>   obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
>   obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
>   obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
> +obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
>
>   # LED SPI Drivers
>   obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
> new file mode 100644
> index 0000000..49818f0
> --- /dev/null
> +++ b/drivers/leds/leds-is31fl32xx.c
> @@ -0,0 +1,505 @@
> +/*
> + * linux/drivers/leds-is31fl32xx.c
> + *
> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
> + *
> + * Copyright 2015 Allworx Corp.
> + *
> + *
> + * 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.
> + *
> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> + */
> +
> +#include <linux/err.h>
> +#include <linux/i2c.h>
> +#include <linux/kernel.h>
> +#include <linux/leds.h>
> +#include <linux/module.h>
> +#include <linux/of_platform.h>
> +
> +/* Used to indicate a device has no such register */
> +#define IS31FL32XX_REG_NONE 0xFF
> +
> +/* Software Shutdown bit in Shutdown Register */
> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
> +
> +/* IS31FL3216 has a number of unique registers */
> +#define IS31FL3216_CONFIG_REG 0x00
> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
> +
> +/* Software Shutdown bit in 3216 Config Register */
> +#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
> +
> +struct is31fl32xx_priv;
> +struct is31fl32xx_led_data {
> +	struct led_classdev cdev;
> +	u8 channel; /* 1-based, max priv->cdef->channels */
> +	struct is31fl32xx_priv *priv;
> +};
> +
> +struct is31fl32xx_priv {
> +	const struct is31fl32xx_chipdef *cdef;
> +	struct i2c_client *client;
> +	unsigned int num_leds;
> +	struct is31fl32xx_led_data leds[0];
> +};
> +
> +/**
> + * struct is31fl32xx_chipdef - chip-specific attributes
> + * @channels            : Number of LED channels
> + * @shutdown_reg        : address of Shutdown register (optional)
> + * @pwm_update_reg      : address of PWM Update register
> + * @global_control_reg  : address of Global Control register (optional)
> + * @reset_reg           : address of Reset register (optional)
> + * @pwm_register_base   : address of first PWM register
> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
> + * @led_control_register_base : address of first LED control register (optional)
> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
> + * @reset_func:         : pointer to reset function
> + *
> + * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
> + * indicates that this chip has no such register.
> + *
> + * If non-NULL, @reset_func will be called during probing to set all
> + * necessary registers to a known initialization state. This is needed
> + * for chips that do not have a @reset_reg.
> + *
> + * @enable_bits_per_led_control_register must be >=1 if
> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
> + */
> +struct is31fl32xx_chipdef {
> +	u8	channels;
> +	u8	shutdown_reg;
> +	u8	pwm_update_reg;
> +	u8	global_control_reg;
> +	u8	reset_reg;
> +	u8	pwm_register_base;
> +	bool	pwm_registers_reversed;
> +	u8	led_control_register_base;
> +	u8	enable_bits_per_led_control_register;
> +	int (*reset_func)(struct is31fl32xx_priv *priv);
> +	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
> +};
> +
> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
> +	.channels				= 36,
> +	.shutdown_reg				= 0x00,
> +	.pwm_update_reg				= 0x25,
> +	.global_control_reg			= 0x4a,
> +	.reset_reg				= 0x4f,
> +	.pwm_register_base			= 0x01,
> +	.led_control_register_base		= 0x26,
> +	.enable_bits_per_led_control_register	= 1,
> +};
> +
> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
> +	.channels				= 28,
> +	.shutdown_reg				= 0x00,
> +	.pwm_update_reg				= 0x25,
> +	.global_control_reg			= 0x4a,
> +	.reset_reg				= 0x4f,
> +	.pwm_register_base			= 0x05,
> +	.led_control_register_base		= 0x2a,
> +	.enable_bits_per_led_control_register	= 1,
> +};
> +
> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
> +	.channels				= 18,
> +	.shutdown_reg				= 0x00,
> +	.pwm_update_reg				= 0x16,
> +	.global_control_reg			= IS31FL32XX_REG_NONE,
> +	.reset_reg				= 0x17,
> +	.pwm_register_base			= 0x01,
> +	.led_control_register_base		= 0x13,
> +	.enable_bits_per_led_control_register	= 6,
> +};
> +
> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> +					bool enable);
> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
> +	.channels				= 16,
> +	.shutdown_reg				= IS31FL32XX_REG_NONE,
> +	.pwm_update_reg				= 0xB0,
> +	.global_control_reg			= IS31FL32XX_REG_NONE,
> +	.reset_reg				= IS31FL32XX_REG_NONE,
> +	.pwm_register_base			= 0x10,
> +	.pwm_registers_reversed			= true,
> +	.led_control_register_base		= 0x01,
> +	.enable_bits_per_led_control_register	= 8,
> +	.reset_func				= is31fl3216_reset,
> +	.sw_shutdown_func			= is31fl3216_software_shutdown,
> +};
> +
> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
> +{
> +	int ret;
> +
> +	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
> +
> +	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
> +	if (ret) {
> +		dev_err(&priv->client->dev,
> +			"register write to 0x%02X failed (error %d)",
> +			reg, ret);
> +	}
> +	return ret;
> +}
> +
> +/*
> + * Custom reset function for IS31FL3216 because it does not have a RESET
> + * register the way that the other IS31FL32xx chips do. We don't bother
> + * writing the GPIO and animation registers, because the registers we
> + * do write ensure those will have no effect.
> + */
> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
> +{
> +	unsigned int i;
> +	int ret;
> +
> +	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
> +			       IS31FL3216_CONFIG_SSD_ENABLE);
> +	if (ret)
> +		return ret;
> +	for (i = 0; i < priv->cdef->channels; i++) {
> +		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
> +				       0x00);
> +		if (ret)
> +			return ret;
> +	}
> +	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
> +	if (ret)
> +		return ret;
> +	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
> +	if (ret)
> +		return ret;
> +	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
> +	if (ret)
> +		return ret;
> +
> +	return 0;
> +}
> +
> +/*
> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
> + * We don't bother doing a read/modify/write on the CONFIG register because
> + * we only ever use a value of '0' for the other fields in that register.
> + */
> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> +					bool enable)
> +{
> +	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
> +			    IS31FL3216_CONFIG_SSD_DISABLE;
> +
> +	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
> +}
> +
> +/*
> + * NOTE: A mutex is not needed in this function because:
> + * - All referenced data is read-only after probe()
> + * - The I2C core has a mutex on to protect the bus
> + * - There are no read/modify/write operations
> + * - Intervening operations between the write of the PWM register
> + *   and the Update register are harmless.
> + *
> + * Example:
> + *	PWM_REG_1 write 16
> + *	UPDATE_REG write 0
> + *	PWM_REG_2 write 128
> + *	UPDATE_REG write 0
> + *   vs:
> + *	PWM_REG_1 write 16
> + *	PWM_REG_2 write 128
> + *	UPDATE_REG write 0
> + *	UPDATE_REG write 0
> + * are equivalent. Poking the Update register merely applies all PWM
> + * register writes up to that point.
> + */
> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
> +				     enum led_brightness brightness)
> +{
> +	const struct is31fl32xx_led_data *led_data =
> +		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
> +	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
> +	u8 pwm_register_offset;
> +	int ret;
> +
> +	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
> +
> +	/* NOTE: led_data->channel is 1-based */
> +	if (cdef->pwm_registers_reversed)
> +		pwm_register_offset = cdef->channels - led_data->channel;
> +	else
> +		pwm_register_offset = led_data->channel - 1;
> +
> +	ret = is31fl32xx_write(led_data->priv,
> +			       cdef->pwm_register_base + pwm_register_offset,
> +			       brightness);
> +	if (ret)
> +		return ret;
> +
> +	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
> +}
> +
> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
> +{
> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> +	int ret;
> +
> +	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
> +		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	if (cdef->reset_func)
> +		return cdef->reset_func(priv);
> +
> +	return 0;
> +}
> +
> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
> +					bool enable)
> +{
> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> +	int ret;
> +
> +	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
> +		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
> +				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
> +		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	if (cdef->sw_shutdown_func)
> +		return cdef->sw_shutdown_func(priv, enable);

You seem to call sw_shutdown_func only here, so why should we have
enable parameter in this op?

> +	return 0;
> +}
> +
> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
> +{
> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> +	int ret;
> +
> +	ret = is31fl32xx_reset_regs(priv);
> +	if (ret)
> +		return ret;
> +
> +	/*
> +	 * Set enable bit for all channels.
> +	 * We will control state with PWM registers alone.
> +	 */
> +	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
> +		u8 value =
> +		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
> +		u8 num_regs = cdef->channels /
> +				cdef->enable_bits_per_led_control_register;
> +		int i;
> +
> +		for (i = 0; i < num_regs; i++) {
> +			ret = is31fl32xx_write(priv,
> +					       cdef->led_control_register_base+i,
> +					       value);
> +			if (ret)
> +				return ret;
> +		}
> +	}
> +
> +	ret = is31fl32xx_software_shutdown(priv, false);
> +	if (ret)
> +		return ret;
> +
> +	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
> +		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	return 0;
> +}
> +
> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
> +{
> +	return sizeof(struct is31fl32xx_priv) +
> +		      (sizeof(struct is31fl32xx_led_data) * num_leds);
> +}
> +
> +static int is31fl32xx_parse_child_dt(const struct device *dev,
> +				     const struct device_node *child,
> +				     struct is31fl32xx_led_data *led_data)
> +{
> +	struct led_classdev *cdev = &led_data->cdev;
> +	int ret = 0;
> +	u32 reg;
> +
> +	if (of_property_read_string(child, "label", &cdev->name))
> +		cdev->name = child->name;
> +
> +	ret = of_property_read_u32(child, "reg", &reg);
> +	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
> +		dev_err(dev,
> +			"Child node %s does not have a valid reg property\n",
> +			child->full_name);
> +		return -EINVAL;
> +	}
> +	led_data->channel = reg;
> +
> +	of_property_read_string(child, "linux,default-trigger",
> +				&cdev->default_trigger);
> +
> +	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
> +
> +	return 0;
> +}
> +
> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
> +					struct is31fl32xx_priv *priv,
> +					u8 channel)
> +{
> +	size_t i;
> +
> +	for (i = 0; i < priv->num_leds; i++) {
> +		if (priv->leds[i].channel == channel)
> +			return &priv->leds[i];
> +	}
> +
> +	return NULL;
> +}
> +
> +static int is31fl32xx_parse_dt(struct device *dev,
> +			       struct is31fl32xx_priv *priv)
> +{
> +	struct device_node *child;
> +	int ret = 0;
> +
> +	for_each_child_of_node(dev->of_node, child) {
> +		struct is31fl32xx_led_data *led_data =
> +			&priv->leds[priv->num_leds];
> +		const struct is31fl32xx_led_data *other_led_data;
> +
> +		led_data->priv = priv;
> +
> +		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
> +		if (ret)
> +			goto err;
> +
> +		/* Detect if channel is already in use by another child */
> +		other_led_data = is31fl32xx_find_led_data(priv,
> +							  led_data->channel);
> +		if (other_led_data) {
> +			dev_err(dev,
> +				"%s and %s both attempting to use channel %d\n",
> +				led_data->cdev.name,
> +				other_led_data->cdev.name,
> +				led_data->channel);
> +			goto err;
> +		}
> +
> +		ret = devm_led_classdev_register(dev, &led_data->cdev);
> +		if (ret) {
> +			dev_err(dev, "failed to register PWM led for %s: %d\n",
> +				led_data->cdev.name, ret);
> +			goto err;
> +		}
> +
> +		priv->num_leds++;
> +	}
> +
> +	return 0;
> +
> +err:
> +	of_node_put(child);
> +	return ret;
> +}
> +
> +static const struct of_device_id of_is31fl31xx_match[] = {
> +	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
> +	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
> +	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
> +	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
> +	{},
> +};
> +
> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
> +
> +static int is31fl32xx_probe(struct i2c_client *client,
> +			    const struct i2c_device_id *id)
> +{
> +	const struct is31fl32xx_chipdef *cdef;
> +	const struct of_device_id *of_dev_id;
> +	struct device *dev = &client->dev;
> +	struct is31fl32xx_priv *priv;
> +	int count;
> +	int ret = 0;
> +
> +	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
> +	if (!of_dev_id)
> +		return -EINVAL;
> +
> +	cdef = of_dev_id->data;
> +
> +	count = of_get_child_count(dev->of_node);
> +	if (!count)
> +		return -EINVAL;
> +
> +	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
> +			    GFP_KERNEL);
> +	if (!priv)
> +		return -ENOMEM;
> +
> +	priv->client = client;
> +	priv->cdef = cdef;
> +	i2c_set_clientdata(client, priv);
> +
> +	ret = is31fl32xx_init_regs(priv);
> +	if (ret)
> +		return ret;
> +
> +	ret = is31fl32xx_parse_dt(dev, priv);
> +	if (ret)
> +		return ret;
> +
> +	return 0;
> +}
> +
> +static int is31fl32xx_remove(struct i2c_client *client)
> +{
> +	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
> +
> +	return is31fl32xx_reset_regs(priv);
> +}
> +
> +/*
> + * i2c-core requires that id_table be non-NULL, even though
> + * it is not used for DeviceTree based instantiation.
> + */
> +static const struct i2c_device_id is31fl31xx_id[] = {
> +	{},
> +};
> +
> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
> +
> +static struct i2c_driver is31fl32xx_driver = {
> +	.driver = {
> +		.name	= "is31fl32xx",
> +		.of_match_table = of_is31fl31xx_match,
> +	},
> +	.probe		= is31fl32xx_probe,
> +	.remove		= is31fl32xx_remove,
> +	.id_table	= is31fl31xx_id,
> +};
> +
> +module_i2c_driver(is31fl32xx_driver);
> +
> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
> +MODULE_LICENSE("GPL v2");
>


-- 
Best regards,
Jacek Anaszewski

Re: [PATCH 4/4] leds: Replace dedicated SN3218 driver with IS31FL32XX driver

[Jacek Anaszewski]

Hi David,

I'll wait for Tested-by from Stefan before applying this patch.
If Stefan will have managed to test your driver with his hardware
by the end of this cycle, it will suffice for this patch to contain
only leds-is31fl32xx extension part.

leds-sn3218 hasn't been merged to mainline yet, so I'd rather
remove it when I get Tested-by from Stefan.

Otherwise, I will send leds-sn3218 to Linus, and this patch
will be applied after being tested, during 4.7 cycle.

Thanks,
Jacek Anaszewski

On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:
> From: David Rivshin <drivshin@allworx.com>
>
> Si-En Technology was acquired by ISSI in 2011, and it appears that
> the IS31FL3218/IS31FL3216 are just rebranded SN3218/SN3216 devices.
> As the IS31FL32XX driver already handles the *3218 devices, there
> is no longer a need for the dedicated SN3218 driver.
>
> Add the "sn,sn3218" and "sn,sn3216" compatible strings into the
> IS31FL32XX driver and binding documentation, and remove the
> leds-sn3218 driver.
>
> Datasheets:
>      IS31FL3218: http://www.issi.com/WW/pdf/31FL3218.pdf
>      SN3218:     http://www.si-en.com/uploadpdf/s2011517171720.pdf
>
>      IS31FL3216: http://www.issi.com/WW/pdf/31FL3216.pdf
>      SN3216;     http://www.si-en.com/uploadpdf/SN3216201152410148.pdf
>
> Signed-off-by: David Rivshin <drivshin@allworx.com>
> ---
>
> Note that the leds-sn3218 binding use a 0-based 'reg' property, while
> the leds-is31fl32xx binding uses a 1-based 'reg' property. This seemed
> to be the preferred binding based on [1]. Since leds-sn3216 has not been
> in a released kernel, there is are no backwards-compatibility concerns.
>
> Changes from RFC:
>   new
>
> [1] http://www.spinics.net/lists/linux-leds/msg05589.html
>
>   .../devicetree/bindings/leds/leds-is31fl32xx.txt   |   9 +-
>   .../devicetree/bindings/leds/leds-sn3218.txt       |  41 ---
>   drivers/leds/Kconfig                               |  18 +-
>   drivers/leds/Makefile                              |   1 -
>   drivers/leds/leds-is31fl32xx.c                     |   6 +-
>   drivers/leds/leds-sn3218.c                         | 306 ---------------------
>   6 files changed, 14 insertions(+), 367 deletions(-)
>   delete mode 100644 Documentation/devicetree/bindings/leds/leds-sn3218.txt
>   delete mode 100644 drivers/leds/leds-sn3218.c
>
> diff --git a/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
> index 539df2e..c59eb1a 100644
> --- a/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
> +++ b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
> @@ -1,6 +1,6 @@
> -Binding for ISSI IS31FL32xx LED Drivers
> +Binding for ISSI IS31FL32xx and Si-En SN32xx LED Drivers
>
> -The IS31FL32xx family of LED drivers are I2C devices with multiple
> +The IS31FL32xx/SN32xx family of LED drivers are I2C devices with multiple
>   constant-current channels, each with independent 256-level PWM control.
>   Each LED is represented as a sub-node of the device.
>
> @@ -10,6 +10,8 @@ Required properties:
>   	issi,is31fl3235
>   	issi,is31fl3218
>   	issi,is31fl3216
> +	si-en,sn3218
> +	si-en,sn3216
>   - reg: I2C slave address
>   - address-cells : must be 1
>   - size-cells : must be 0
> @@ -45,5 +47,6 @@ leds: is31fl3236@3c {
>   	};
>   };
>
> -For more product information please see the link below:
> +For more product information please see the links below:
>   http://www.issi.com/US/product-analog-fxled-driver.shtml
> +http://www.si-en.com/product.asp?parentid=890
> diff --git a/Documentation/devicetree/bindings/leds/leds-sn3218.txt b/Documentation/devicetree/bindings/leds/leds-sn3218.txt
> deleted file mode 100644
> index 19cbf57..0000000
> --- a/Documentation/devicetree/bindings/leds/leds-sn3218.txt
> +++ /dev/null
> @@ -1,41 +0,0 @@
> -* Si-En Technology - SN3218 18-Channel LED Driver
> -
> -Required properties:
> -- compatible :
> -	"si-en,sn3218"
> -- address-cells : must be 1
> -- size-cells : must be 0
> -- reg : I2C slave address, typically 0x54
> -
> -There must be at least 1 LED which is represented as a sub-node
> -of the sn3218 device.
> -
> -LED sub-node properties:
> -- label : (optional) see Documentation/devicetree/bindings/leds/common.txt
> -- reg : number of LED line (could be from 0 to 17)
> -- linux,default-trigger : (optional)
> -   see Documentation/devicetree/bindings/leds/common.txt
> -
> -Example:
> -
> -sn3218: led-controller@54 {
> -	compatible = "si-en,sn3218";
> -	#address-cells = <1>;
> -	#size-cells = <0>;
> -	reg = <0x54>;
> -
> -	led@0 {
> -		label = "led1";
> -		reg = <0x0>;
> -	};
> -
> -	led@1 {
> -		label = "led2";
> -		reg = <0x1>;
> -	};
> -
> -	led@2 {
> -		label = "led3";
> -		reg = <0x2>;
> -	};
> -};
> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> index 9c63ba4..1f64151 100644
> --- a/drivers/leds/Kconfig
> +++ b/drivers/leds/Kconfig
> @@ -568,25 +568,13 @@ config LEDS_SEAD3
>   	  This driver can also be built as a module. If so the module
>   	  will be called leds-sead3.
>
> -config LEDS_SN3218
> -	tristate "LED support for Si-En SN3218 I2C chip"
> -	depends on LEDS_CLASS && I2C
> -	depends on OF
> -	select REGMAP_I2C
> -	help
> -	  This option enables support for the Si-EN SN3218 LED driver
> -	  connected through I2C. Say Y to enable support for the SN3218 LED.
> -
> -	  This driver can also be built as a module. If so the module
> -	  will be called leds-sn3218.
> -
>   config LEDS_IS31FL32XX
>   	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
>   	depends on LEDS_CLASS && I2C && OF
>   	help
> -	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
> -	  They are I2C devices with multiple constant-current channels, each
> -	  with independent 256-level PWM control.
> +	  Say Y here to include support for ISSI IS31FL32XX and Si-En SN32xx
> +	  LED controllers. They are I2C devices with multiple constant-current
> +	  channels, each with independent 256-level PWM control.
>
>   comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
>
> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
> index 3fdf313..cb2013d 100644
> --- a/drivers/leds/Makefile
> +++ b/drivers/leds/Makefile
> @@ -66,7 +66,6 @@ obj-$(CONFIG_LEDS_MENF21BMC)		+= leds-menf21bmc.o
>   obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
>   obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
>   obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
> -obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
>   obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
>
>   # LED SPI Drivers
> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
> index 49818f0..ec3f541 100644
> --- a/drivers/leds/leds-is31fl32xx.c
> +++ b/drivers/leds/leds-is31fl32xx.c
> @@ -10,7 +10,9 @@
>    * it under the terms of the GNU General Public License version 2 as
>    * published by the Free Software Foundation.
>    *
> - * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> + * Datasheets:
> + *   http://www.issi.com/US/product-analog-fxled-driver.shtml
> + *   http://www.si-en.com/product.asp?parentid=890
>    */
>
>   #include <linux/err.h>
> @@ -425,7 +427,9 @@ static const struct of_device_id of_is31fl31xx_match[] = {
>   	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
>   	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
>   	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
> +	{ .compatible = "si-en,sn3218",    .data = &is31fl3218_cdef, },
>   	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
> +	{ .compatible = "si-en,sn3216",    .data = &is31fl3216_cdef, },
>   	{},
>   };
>
> diff --git a/drivers/leds/leds-sn3218.c b/drivers/leds/leds-sn3218.c
> deleted file mode 100644
> index dcc2581..0000000
> --- a/drivers/leds/leds-sn3218.c
> +++ /dev/null
> @@ -1,306 +0,0 @@
> -/*
> - * Si-En SN3218 18 Channel LED Driver
> - *
> - * Copyright (C) 2016 Stefan Wahren <stefan.wahren@i2se.com>
> - *
> - * Based on leds-pca963x.c
> - *
> - * 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.
> - *
> - * Datasheet: http://www.si-en.com/uploadpdf/s2011517171720.pdf
> - *
> - */
> -
> -#include <linux/err.h>
> -#include <linux/i2c.h>
> -#include <linux/leds.h>
> -#include <linux/module.h>
> -#include <linux/of.h>
> -#include <linux/regmap.h>
> -#include <linux/slab.h>
> -
> -#define SN3218_MODE		0x00
> -#define SN3218_PWM_1		0x01
> -#define SN3218_PWM_2		0x02
> -#define SN3218_PWM_3		0x03
> -#define SN3218_PWM_4		0x04
> -#define SN3218_PWM_5		0x05
> -#define SN3218_PWM_6		0x06
> -#define SN3218_PWM_7		0x07
> -#define SN3218_PWM_8		0x08
> -#define SN3218_PWM_9		0x09
> -#define SN3218_PWM_10		0x0a
> -#define SN3218_PWM_11		0x0b
> -#define SN3218_PWM_12		0x0c
> -#define SN3218_PWM_13		0x0d
> -#define SN3218_PWM_14		0x0e
> -#define SN3218_PWM_15		0x0f
> -#define SN3218_PWM_16		0x10
> -#define SN3218_PWM_17		0x11
> -#define SN3218_PWM_18		0x12
> -#define SN3218_LED_1_6		0x13
> -#define SN3218_LED_7_12		0x14
> -#define SN3218_LED_13_18	0x15
> -#define SN3218_UPDATE		0x16	/* applies to reg 0x01 .. 0x15 */
> -#define SN3218_RESET		0x17
> -
> -#define SN3218_LED_MASK		0x3f
> -#define SN3218_LED_ON		0x01
> -#define SN3218_LED_OFF		0x00
> -
> -#define SN3218_MODE_SHUTDOWN	0x00
> -#define SN3218_MODE_NORMAL	0x01
> -
> -#define NUM_LEDS		18
> -
> -struct sn3218_led;
> -
> -/**
> - * struct sn3218 -
> - * @client - Pointer to the I2C client
> - * @leds - Pointer to the individual LEDs
> - * @num_leds - Actual number of LEDs
> -**/
> -struct sn3218 {
> -	struct i2c_client *client;
> -	struct regmap *regmap;
> -	struct sn3218_led *leds;
> -	int num_leds;
> -};
> -
> -/**
> - * struct sn3218_led -
> - * @chip - Pointer to the container
> - * @led_cdev - led class device pointer
> - * @led_num - LED index ( 0 .. 17 )
> -**/
> -struct sn3218_led {
> -	struct sn3218 *chip;
> -	struct led_classdev led_cdev;
> -	int led_num;
> -};
> -
> -static int sn3218_led_set(struct led_classdev *led_cdev,
> -			  enum led_brightness brightness)
> -{
> -	struct sn3218_led *led =
> -			container_of(led_cdev, struct sn3218_led, led_cdev);
> -	struct regmap *regmap = led->chip->regmap;
> -	u8 bank = led->led_num / 6;
> -	u8 mask = 0x1 << (led->led_num % 6);
> -	u8 val;
> -	int ret;
> -
> -	if (brightness == LED_OFF)
> -		val = 0;
> -	else
> -		val = mask;
> -
> -	ret = regmap_update_bits(regmap, SN3218_LED_1_6 + bank, mask, val);
> -	if (ret < 0)
> -		return ret;
> -
> -	if (brightness > LED_OFF) {
> -		ret = regmap_write(regmap, SN3218_PWM_1 + led->led_num,
> -				   brightness);
> -		if (ret < 0)
> -			return ret;
> -	}
> -
> -	ret = regmap_write(regmap, SN3218_UPDATE, 0xff);
> -
> -	return ret;
> -}
> -
> -static void sn3218_led_init(struct sn3218 *sn3218, struct device_node *node,
> -			    u32 reg)
> -{
> -	struct sn3218_led *leds = sn3218->leds;
> -	struct led_classdev *cdev = &leds[reg].led_cdev;
> -
> -	leds[reg].led_num = reg;
> -	leds[reg].chip = sn3218;
> -
> -	if (of_property_read_string(node, "label", &cdev->name))
> -		cdev->name = node->name;
> -
> -	of_property_read_string(node, "linux,default-trigger",
> -				&cdev->default_trigger);
> -
> -	cdev->brightness_set_blocking = sn3218_led_set;
> -}
> -
> -static const struct reg_default sn3218_reg_defs[] = {
> -	{ SN3218_MODE, 0x00},
> -	{ SN3218_PWM_1, 0x00},
> -	{ SN3218_PWM_2, 0x00},
> -	{ SN3218_PWM_3, 0x00},
> -	{ SN3218_PWM_4, 0x00},
> -	{ SN3218_PWM_5, 0x00},
> -	{ SN3218_PWM_6, 0x00},
> -	{ SN3218_PWM_7, 0x00},
> -	{ SN3218_PWM_8, 0x00},
> -	{ SN3218_PWM_9, 0x00},
> -	{ SN3218_PWM_10, 0x00},
> -	{ SN3218_PWM_11, 0x00},
> -	{ SN3218_PWM_12, 0x00},
> -	{ SN3218_PWM_13, 0x00},
> -	{ SN3218_PWM_14, 0x00},
> -	{ SN3218_PWM_15, 0x00},
> -	{ SN3218_PWM_16, 0x00},
> -	{ SN3218_PWM_17, 0x00},
> -	{ SN3218_PWM_18, 0x00},
> -	{ SN3218_LED_1_6, 0x00},
> -	{ SN3218_LED_7_12, 0x00},
> -	{ SN3218_LED_13_18, 0x00},
> -	{ SN3218_UPDATE, 0x00},
> -	{ SN3218_RESET, 0x00},
> -};
> -
> -static const struct regmap_config sn3218_regmap_config = {
> -	.reg_bits = 8,
> -	.val_bits = 8,
> -
> -	.max_register = SN3218_RESET,
> -	.reg_defaults = sn3218_reg_defs,
> -	.num_reg_defaults = ARRAY_SIZE(sn3218_reg_defs),
> -	.cache_type = REGCACHE_RBTREE,
> -};
> -
> -static int sn3218_init(struct i2c_client *client, struct sn3218 *sn3218)
> -{
> -	struct device_node *np = client->dev.of_node, *child;
> -	struct sn3218_led *leds;
> -	int ret, count;
> -
> -	count = of_get_child_count(np);
> -	if (!count)
> -		return -ENODEV;
> -
> -	if (count > NUM_LEDS) {
> -		dev_err(&client->dev, "Invalid LED count %d\n", count);
> -		return -EINVAL;
> -	}
> -
> -	leds = devm_kzalloc(&client->dev, count * sizeof(*leds), GFP_KERNEL);
> -	if (!leds)
> -		return -ENOMEM;
> -
> -	sn3218->leds = leds;
> -	sn3218->num_leds = count;
> -	sn3218->client = client;
> -
> -	sn3218->regmap = devm_regmap_init_i2c(client, &sn3218_regmap_config);
> -	if (IS_ERR(sn3218->regmap)) {
> -		ret = PTR_ERR(sn3218->regmap);
> -		dev_err(&client->dev, "Failed to allocate register map: %d\n",
> -			ret);
> -		return ret;
> -	}
> -
> -	for_each_child_of_node(np, child) {
> -		u32 reg;
> -
> -		ret = of_property_read_u32(child, "reg", &reg);
> -		if (ret)
> -			goto fail;
> -
> -		if (reg >= count) {
> -			dev_err(&client->dev, "Invalid LED (%u >= %d)\n", reg,
> -				count);
> -			ret = -EINVAL;
> -			goto fail;
> -		}
> -
> -		sn3218_led_init(sn3218, child, reg);
> -	}
> -
> -	return 0;
> -
> -fail:
> -	of_node_put(child);
> -	return ret;
> -}
> -
> -static int sn3218_probe(struct i2c_client *client,
> -			const struct i2c_device_id *id)
> -{
> -	struct sn3218 *sn3218;
> -	struct sn3218_led *leds;
> -	struct device *dev = &client->dev;
> -	int i, ret;
> -
> -	sn3218 = devm_kzalloc(dev, sizeof(*sn3218), GFP_KERNEL);
> -	if (!sn3218)
> -		return -ENOMEM;
> -
> -	ret = sn3218_init(client, sn3218);
> -	if (ret)
> -		return ret;
> -
> -	i2c_set_clientdata(client, sn3218);
> -	leds = sn3218->leds;
> -
> -	/*
> -	 * Since the chip is write-only we need to reset him into
> -	 * a defined state (all LEDs off).
> -	 */
> -	ret = regmap_write(sn3218->regmap, SN3218_RESET, 0xff);
> -	if (ret)
> -		return ret;
> -
> -	for (i = 0; i < sn3218->num_leds; i++) {
> -		ret = devm_led_classdev_register(dev, &leds[i].led_cdev);
> -		if (ret < 0)
> -			return ret;
> -	}
> -
> -	return regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_NORMAL);
> -}
> -
> -static int sn3218_remove(struct i2c_client *client)
> -{
> -	struct sn3218 *sn3218 = i2c_get_clientdata(client);
> -
> -	regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_SHUTDOWN);
> -
> -	return 0;
> -}
> -
> -static void sn3218_shutdown(struct i2c_client *client)
> -{
> -	struct sn3218 *sn3218 = i2c_get_clientdata(client);
> -
> -	regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_SHUTDOWN);
> -}
> -
> -static const struct i2c_device_id sn3218_id[] = {
> -	{ "sn3218", 0 },
> -	{ }
> -};
> -MODULE_DEVICE_TABLE(i2c, sn3218_id);
> -
> -static const struct of_device_id of_sn3218_match[] = {
> -	{ .compatible = "si-en,sn3218", },
> -	{},
> -};
> -MODULE_DEVICE_TABLE(of, of_sn3218_match);
> -
> -static struct i2c_driver sn3218_driver = {
> -	.driver = {
> -		.name	= "leds-sn3218",
> -		.of_match_table = of_match_ptr(of_sn3218_match),
> -	},
> -	.probe	= sn3218_probe,
> -	.remove	= sn3218_remove,
> -	.shutdown = sn3218_shutdown,
> -	.id_table = sn3218_id,
> -};
> -
> -module_i2c_driver(sn3218_driver);
> -
> -MODULE_DESCRIPTION("Si-En SN3218 LED Driver");
> -MODULE_AUTHOR("Stefan Wahren <stefan.wahren@i2se.com>");
> -MODULE_LICENSE("GPL v2");
>

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[Stefan Wahren]

Hi David,

i will test the driver on weekend. Some comments below

> "David Rivshin (Allworx)" <drivshin.allworx@gmail.com> hat am 3. März 2016 um
> 04:01 geschrieben:
>
>
> From: David Rivshin <drivshin@allworx.com>
>
> The IS31FL32xx family of LED controllers are I2C devices with multiple
> constant-current channels, each with independent 256-level PWM control.
>
> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>
> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
> (TI am335x) platform.
>
> The programming paradigm of these devices is similar in the following
> ways:
> - All registers are 8 bit
> - All LED control registers are write-only
> - Each LED channel has a PWM register (0-255)
> - PWM register writes are shadowed until an Update register is poked
> - All have a concept of Software Shutdown, which disables output
>
> However, there are some differences in devices:
> - 3236/3235 have a separate Control register for each LED,
> (3218/3216 pack the enable bits into fewer registers)
> - 3236/3235 have a per-channel current divisor setting
> - 3236/3235 have a Global Control register that can turn off all LEDs
> - 3216 is unique in a number of ways
> - OUT9-OUT16 can be configured as GPIOs instead of LED controls
> - LEDs can be programmed with an 8-frame animation, with
> programmable delay between frames
> - LEDs can be modulated by an input audio signal
> - Max output current can be adjusted from 1/4 to 2x globally
> - Has a Configuration register instead of a Shutdown register
>
> This driver currently only supports the base PWM control function
> of these devices. The following features of these devices are not
> implemented, although it should be possible to add them in the future:
> - All devices are capable of going into a lower-power "software
> shutdown" mode.
> - The is31fl3236 and is31fl3235 can reduce the max output current
> per-channel with a divisor of 1, 2, 3, or 4.
> - The is31fl3216 can use some LED channels as GPIOs instead.
> - The is31fl3216 can animate LEDs in hardware.
> - The is31fl3216 can modulate LEDs according to an audio input.
> - The is31fl3216 can reduce/increase max output current globally.
>
> Signed-off-by: David Rivshin <drivshin@allworx.com>
> ---
>
> You may see two instances of this warning:
> "passing argument 1 of 'of_property_read_string' discards 'const'
> qualifier from pointer target type"
> That is a result of of_property_read_string() taking a non-const
> struct device_node pointer parameter. I have separately submitted a
> patch to fix that [1], and a few related functions which had the same
> issue. I'm hoping that will get into linux-next before this does, so
> that the warnings never show up there.
>
> Changes from RFC:
> - Removed max-brightness DT property.
> - Refer to these devices as "LED controllers" in Kconfig.
> - Removed redundant last sentence from Kconfig entry
> - Removed unnecessary debug code.
> - Do not set led_classdev.brightness to 0 explicitly, as it is
> already initialized to 0 by devm_kzalloc().
> - Used of_property_read_string() instead of of_get_property().
> - Fail immediately on DT parsing error in a child node, rather than
> continuing on with the non-faulty ones.
> - Added additional comments for some things that might be non-obvious.
> - Added constants for the location of the SSD bit in the SHUTDOWN
> register, and the 3216's CONFIG register.
> - Added special sw_shutdown_func for the 3216 device, as that bit
> is in a different register, at a different position, and has reverse
> polarity compared to all the other devices.
> - Refactored is31fl32xx_init_regs() to separate out some logic into
> is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
>
> [1] https://lkml.org/lkml/2016/3/2/746
>
> drivers/leds/Kconfig | 8 +
> drivers/leds/Makefile | 1 +
> drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
> 3 files changed, 514 insertions(+)
> create mode 100644 drivers/leds/leds-is31fl32xx.c
>
> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> index 1034696..9c63ba4 100644
> --- a/drivers/leds/Kconfig
> +++ b/drivers/leds/Kconfig
> @@ -580,6 +580,14 @@ config LEDS_SN3218
> This driver can also be built as a module. If so the module
> will be called leds-sn3218.
>
> +config LEDS_IS31FL32XX
> + tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
> + depends on LEDS_CLASS && I2C && OF
> + help
> + Say Y here to include support for ISSI IS31FL32XX LED controllers.
> + They are I2C devices with multiple constant-current channels, each
> + with independent 256-level PWM control.

Is it worth to mention the module name here?

> +
> comment "LED driver for blink(1) USB RGB LED is under Special HID drivers
> (HID_THINGM)"
>
> config LEDS_BLINKM
> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
> index 89c9b6f..3fdf313 100644
> --- a/drivers/leds/Makefile
> +++ b/drivers/leds/Makefile
> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692) += leds-ktd2692.o
> obj-$(CONFIG_LEDS_POWERNV) += leds-powernv.o
> obj-$(CONFIG_LEDS_SEAD3) += leds-sead3.o
> obj-$(CONFIG_LEDS_SN3218) += leds-sn3218.o
> +obj-$(CONFIG_LEDS_IS31FL32XX) += leds-is31fl32xx.o
>
> # LED SPI Drivers
> obj-$(CONFIG_LEDS_DAC124S085) += leds-dac124s085.o
> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
> new file mode 100644
> index 0000000..49818f0
> --- /dev/null
> +++ b/drivers/leds/leds-is31fl32xx.c
> @@ -0,0 +1,505 @@
> +/*
> + * linux/drivers/leds-is31fl32xx.c

I think this is unnecessary.

> + *
> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
> + *
> + * Copyright 2015 Allworx Corp.
> + *
> + *
> + * 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.
> + *
> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> + */
> +

Shouldn't we include <linux/device.h> here?

> +#include <linux/err.h>
> +#include <linux/i2c.h>
> +#include <linux/kernel.h>
> +#include <linux/leds.h>
> +#include <linux/module.h>
> +#include <linux/of_platform.h>
> +
> +/* Used to indicate a device has no such register */
> +#define IS31FL32XX_REG_NONE 0xFF
> +
> +/* Software Shutdown bit in Shutdown Register */
> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE 0
> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
> +
> +/* IS31FL3216 has a number of unique registers */
> +#define IS31FL3216_CONFIG_REG 0x00
> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
> +
> +/* Software Shutdown bit in 3216 Config Register */
> +#define IS31FL3216_CONFIG_SSD_ENABLE BIT(7)
> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
> +
> +struct is31fl32xx_priv;
> +struct is31fl32xx_led_data {
> + struct led_classdev cdev;
> + u8 channel; /* 1-based, max priv->cdef->channels */
> + struct is31fl32xx_priv *priv;
> +};
> +
> +struct is31fl32xx_priv {
> + const struct is31fl32xx_chipdef *cdef;
> + struct i2c_client *client;
> + unsigned int num_leds;
> + struct is31fl32xx_led_data leds[0];
> +};
> +
> +/**
> + * struct is31fl32xx_chipdef - chip-specific attributes
> + * @channels : Number of LED channels
> + * @shutdown_reg : address of Shutdown register (optional)
> + * @pwm_update_reg : address of PWM Update register
> + * @global_control_reg : address of Global Control register (optional)
> + * @reset_reg : address of Reset register (optional)
> + * @pwm_register_base : address of first PWM register
> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
> + * @led_control_register_base : address of first LED control register
> (optional)
> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
> + * @reset_func: : pointer to reset function
> + *
> + * For all optional register addresses, the sentinel value
> %IS31FL32XX_REG_NONE
> + * indicates that this chip has no such register.
> + *
> + * If non-NULL, @reset_func will be called during probing to set all
> + * necessary registers to a known initialization state. This is needed
> + * for chips that do not have a @reset_reg.
> + *
> + * @enable_bits_per_led_control_register must be >=1 if
> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
> + */
> +struct is31fl32xx_chipdef {
> + u8 channels;
> + u8 shutdown_reg;
> + u8 pwm_update_reg;
> + u8 global_control_reg;
> + u8 reset_reg;
> + u8 pwm_register_base;
> + bool pwm_registers_reversed;
> + u8 led_control_register_base;
> + u8 enable_bits_per_led_control_register;
> + int (*reset_func)(struct is31fl32xx_priv *priv);
> + int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
> +};
> +
> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
> + .channels = 36,
> + .shutdown_reg = 0x00,
> + .pwm_update_reg = 0x25,
> + .global_control_reg = 0x4a,
> + .reset_reg = 0x4f,
> + .pwm_register_base = 0x01,
> + .led_control_register_base = 0x26,
> + .enable_bits_per_led_control_register = 1,
> +};
> +
> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
> + .channels = 28,
> + .shutdown_reg = 0x00,
> + .pwm_update_reg = 0x25,
> + .global_control_reg = 0x4a,
> + .reset_reg = 0x4f,
> + .pwm_register_base = 0x05,
> + .led_control_register_base = 0x2a,
> + .enable_bits_per_led_control_register = 1,
> +};
> +
> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
> + .channels = 18,
> + .shutdown_reg = 0x00,
> + .pwm_update_reg = 0x16,
> + .global_control_reg = IS31FL32XX_REG_NONE,
> + .reset_reg = 0x17,
> + .pwm_register_base = 0x01,
> + .led_control_register_base = 0x13,
> + .enable_bits_per_led_control_register = 6,
> +};
> +
> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> + bool enable);
> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
> + .channels = 16,
> + .shutdown_reg = IS31FL32XX_REG_NONE,
> + .pwm_update_reg = 0xB0,
> + .global_control_reg = IS31FL32XX_REG_NONE,
> + .reset_reg = IS31FL32XX_REG_NONE,
> + .pwm_register_base = 0x10,
> + .pwm_registers_reversed = true,
> + .led_control_register_base = 0x01,
> + .enable_bits_per_led_control_register = 8,
> + .reset_func = is31fl3216_reset,
> + .sw_shutdown_func = is31fl3216_software_shutdown,
> +};
> +
> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
> +{
> + int ret;
> +
> + dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
> +
> + ret = i2c_smbus_write_byte_data(priv->client, reg, val);
> + if (ret) {
> + dev_err(&priv->client->dev,
> + "register write to 0x%02X failed (error %d)",
> + reg, ret);
> + }

In case somebody use this driver as heartbeat and writing fails permanently the
log will be flooded.

> + return ret;
> +}
> +
> +/*
> + * Custom reset function for IS31FL3216 because it does not have a RESET
> + * register the way that the other IS31FL32xx chips do. We don't bother
> + * writing the GPIO and animation registers, because the registers we
> + * do write ensure those will have no effect.
> + */
> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
> +{
> + unsigned int i;
> + int ret;
> +
> + ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
> + IS31FL3216_CONFIG_SSD_ENABLE);
> + if (ret)
> + return ret;
> + for (i = 0; i < priv->cdef->channels; i++) {
> + ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
> + 0x00);
> + if (ret)
> + return ret;
> + }
> + ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
> + if (ret)
> + return ret;
> + ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
> + if (ret)
> + return ret;
> + ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
> + if (ret)
> + return ret;
> +
> + return 0;
> +}
> +
> +/*
> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
> + * We don't bother doing a read/modify/write on the CONFIG register because
> + * we only ever use a value of '0' for the other fields in that register.
> + */
> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> + bool enable)
> +{
> + u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
> + IS31FL3216_CONFIG_SSD_DISABLE;
> +
> + return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
> +}
> +
> +/*
> + * NOTE: A mutex is not needed in this function because:
> + * - All referenced data is read-only after probe()
> + * - The I2C core has a mutex on to protect the bus
> + * - There are no read/modify/write operations
> + * - Intervening operations between the write of the PWM register
> + * and the Update register are harmless.
> + *
> + * Example:
> + * PWM_REG_1 write 16
> + * UPDATE_REG write 0
> + * PWM_REG_2 write 128
> + * UPDATE_REG write 0
> + * vs:
> + * PWM_REG_1 write 16
> + * PWM_REG_2 write 128
> + * UPDATE_REG write 0
> + * UPDATE_REG write 0
> + * are equivalent. Poking the Update register merely applies all PWM
> + * register writes up to that point.
> + */
> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
> + enum led_brightness brightness)
> +{
> + const struct is31fl32xx_led_data *led_data =
> + container_of(led_cdev, struct is31fl32xx_led_data, cdev);
> + const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
> + u8 pwm_register_offset;
> + int ret;
> +
> + dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
> +
> + /* NOTE: led_data->channel is 1-based */
> + if (cdef->pwm_registers_reversed)
> + pwm_register_offset = cdef->channels - led_data->channel;
> + else
> + pwm_register_offset = led_data->channel - 1;
> +
> + ret = is31fl32xx_write(led_data->priv,
> + cdef->pwm_register_base + pwm_register_offset,
> + brightness);
> + if (ret)
> + return ret;
> +
> + return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
> +}
> +
> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
> +{
> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
> + int ret;
> +
> + if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
> + ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
> + if (ret)
> + return ret;
> + }
> +
> + if (cdef->reset_func)
> + return cdef->reset_func(priv);
> +
> + return 0;
> +}
> +
> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
> + bool enable)
> +{
> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
> + int ret;
> +
> + if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
> + u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
> + IS31FL32XX_SHUTDOWN_SSD_DISABLE;
> + ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
> + if (ret)
> + return ret;
> + }
> +
> + if (cdef->sw_shutdown_func)
> + return cdef->sw_shutdown_func(priv, enable);
> +
> + return 0;
> +}
> +
> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
> +{
> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
> + int ret;
> +
> + ret = is31fl32xx_reset_regs(priv);
> + if (ret)
> + return ret;
> +
> + /*
> + * Set enable bit for all channels.
> + * We will control state with PWM registers alone.
> + */
> + if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
> + u8 value =
> + GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
> + u8 num_regs = cdef->channels /
> + cdef->enable_bits_per_led_control_register;
> + int i;
> +
> + for (i = 0; i < num_regs; i++) {
> + ret = is31fl32xx_write(priv,
> + cdef->led_control_register_base+i,
> + value);
> + if (ret)
> + return ret;
> + }
> + }
> +
> + ret = is31fl32xx_software_shutdown(priv, false);
> + if (ret)
> + return ret;
> +
> + if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
> + ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
> + if (ret)
> + return ret;
> + }
> +
> + return 0;
> +}
> +
> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
> +{
> + return sizeof(struct is31fl32xx_priv) +
> + (sizeof(struct is31fl32xx_led_data) * num_leds);
> +}
> +
> +static int is31fl32xx_parse_child_dt(const struct device *dev,
> + const struct device_node *child,
> + struct is31fl32xx_led_data *led_data)
> +{
> + struct led_classdev *cdev = &led_data->cdev;
> + int ret = 0;
> + u32 reg;
> +
> + if (of_property_read_string(child, "label", &cdev->name))
> + cdev->name = child->name;
> +
> + ret = of_property_read_u32(child, "reg", &reg);
> + if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
> + dev_err(dev,
> + "Child node %s does not have a valid reg property\n",
> + child->full_name);
> + return -EINVAL;
> + }
> + led_data->channel = reg;
> +
> + of_property_read_string(child, "linux,default-trigger",
> + &cdev->default_trigger);
> +
> + cdev->brightness_set_blocking = is31fl32xx_brightness_set;
> +
> + return 0;
> +}
> +
> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
> + struct is31fl32xx_priv *priv,
> + u8 channel)
> +{
> + size_t i;
> +
> + for (i = 0; i < priv->num_leds; i++) {
> + if (priv->leds[i].channel == channel)
> + return &priv->leds[i];
> + }
> +
> + return NULL;
> +}
> +
> +static int is31fl32xx_parse_dt(struct device *dev,
> + struct is31fl32xx_priv *priv)
> +{
> + struct device_node *child;
> + int ret = 0;
> +
> + for_each_child_of_node(dev->of_node, child) {
> + struct is31fl32xx_led_data *led_data =
> + &priv->leds[priv->num_leds];

Maybe i missed something, but is it really protected against out of index
access?

> + const struct is31fl32xx_led_data *other_led_data;
> +
> + led_data->priv = priv;
> +
> + ret = is31fl32xx_parse_child_dt(dev, child, led_data);
> + if (ret)
> + goto err;
> +
> + /* Detect if channel is already in use by another child */
> + other_led_data = is31fl32xx_find_led_data(priv,
> + led_data->channel);
> + if (other_led_data) {
> + dev_err(dev,
> + "%s and %s both attempting to use channel %d\n",
> + led_data->cdev.name,
> + other_led_data->cdev.name,
> + led_data->channel);
> + goto err;
> + }
> +
> + ret = devm_led_classdev_register(dev, &led_data->cdev);
> + if (ret) {
> + dev_err(dev, "failed to register PWM led for %s: %d\n",
> + led_data->cdev.name, ret);
> + goto err;
> + }
> +
> + priv->num_leds++;
> + }
> +
> + return 0;
> +
> +err:
> + of_node_put(child);
> + return ret;
> +}
> +
> +static const struct of_device_id of_is31fl31xx_match[] = {
> + { .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
> + { .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
> + { .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
> + { .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
> + {},
> +};
> +
> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
> +
> +static int is31fl32xx_probe(struct i2c_client *client,
> + const struct i2c_device_id *id)
> +{
> + const struct is31fl32xx_chipdef *cdef;
> + const struct of_device_id *of_dev_id;
> + struct device *dev = &client->dev;
> + struct is31fl32xx_priv *priv;
> + int count;
> + int ret = 0;
> +
> + of_dev_id = of_match_device(of_is31fl31xx_match, dev);
> + if (!of_dev_id)
> + return -EINVAL;
> +
> + cdef = of_dev_id->data;
> +
> + count = of_get_child_count(dev->of_node);
> + if (!count)
> + return -EINVAL;
> +
> + priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
> + GFP_KERNEL);
> + if (!priv)
> + return -ENOMEM;
> +
> + priv->client = client;
> + priv->cdef = cdef;
> + i2c_set_clientdata(client, priv);
> +
> + ret = is31fl32xx_init_regs(priv);
> + if (ret)
> + return ret;
> +
> + ret = is31fl32xx_parse_dt(dev, priv);
> + if (ret)
> + return ret;
> +
> + return 0;
> +}
> +
> +static int is31fl32xx_remove(struct i2c_client *client)
> +{
> + struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
> +
> + return is31fl32xx_reset_regs(priv);
> +}
> +
> +/*
> + * i2c-core requires that id_table be non-NULL, even though
> + * it is not used for DeviceTree based instantiation.
> + */
> +static const struct i2c_device_id is31fl31xx_id[] = {
> + {},
> +};
> +
> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
> +
> +static struct i2c_driver is31fl32xx_driver = {
> + .driver = {
> + .name = "is31fl32xx",
> + .of_match_table = of_is31fl31xx_match,
> + },
> + .probe = is31fl32xx_probe,
> + .remove = is31fl32xx_remove,

Sorry, what was the reason to skip shutdown?

Thanks
Stefan

> + .id_table = is31fl31xx_id,
> +};
> +
> +module_i2c_driver(is31fl32xx_driver);
> +
> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
> +MODULE_LICENSE("GPL v2");
> --
> 2.5.0
>

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[David Rivshin (Allworx)]

On Thu, 03 Mar 2016 15:51:32 +0100
Jacek Anaszewski <j.anaszewski@samsung.com> wrote:

> Hi David,
> 
> Thanks for the update. Two remarks in the code.
> 
> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:
> > From: David Rivshin <drivshin@allworx.com>
> >
> > The IS31FL32xx family of LED controllers are I2C devices with multiple
> > constant-current channels, each with independent 256-level PWM control.
> >
> > Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >
> > This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
> > (TI am335x) platform.
> >
> > The programming paradigm of these devices is similar in the following
> > ways:
> >   - All registers are 8 bit
> >   - All LED control registers are write-only
> >   - Each LED channel has a PWM register (0-255)
> >   - PWM register writes are shadowed until an Update register is poked
> >   - All have a concept of Software Shutdown, which disables output
> >
> > However, there are some differences in devices:
> >   - 3236/3235 have a separate Control register for each LED,
> >     (3218/3216 pack the enable bits into fewer registers)
> >   - 3236/3235 have a per-channel current divisor setting
> >   - 3236/3235 have a Global Control register that can turn off all LEDs
> >   - 3216 is unique in a number of ways
> >      - OUT9-OUT16 can be configured as GPIOs instead of LED controls
> >      - LEDs can be programmed with an 8-frame animation, with
> >        programmable delay between frames
> >      - LEDs can be modulated by an input audio signal
> >      - Max output current can be adjusted from 1/4 to 2x globally
> >      - Has a Configuration register instead of a Shutdown register
> >
> > This driver currently only supports the base PWM control function
> > of these devices. The following features of these devices are not
> > implemented, although it should be possible to add them in the future:
> >   - All devices are capable of going into a lower-power "software
> >     shutdown" mode.
> >   - The is31fl3236 and is31fl3235 can reduce the max output current
> >     per-channel with a divisor of 1, 2, 3, or 4.
> >   - The is31fl3216 can use some LED channels as GPIOs instead.
> >   - The is31fl3216 can animate LEDs in hardware.
> >   - The is31fl3216 can modulate LEDs according to an audio input.
> >   - The is31fl3216 can reduce/increase max output current globally.
> >
> > Signed-off-by: David Rivshin <drivshin@allworx.com>
> > ---
> >
> > You may see two instances of this warning:
> >    "passing argument 1 of 'of_property_read_string' discards 'const'
> >     qualifier from pointer target type"
> > That is a result of of_property_read_string() taking a non-const
> > struct device_node pointer parameter. I have separately submitted a
> > patch to fix that [1], and a few related functions which had the same
> > issue. I'm hoping that will get into linux-next before this does, so
> > that the warnings never show up there.  
> 
> Please adjust the patch so that it compiles without warnings on
> current linux-next. Your patch for DT API hasn't been reviewed yet
> AFICS, and I can imagine that there will be some resistance against.

Since the DT API patch was just accepted by Rob [1], would it be OK 
to wait for the results of Stefan's testing (and any other reviews)
before making a decision on this? From Stefan's note, it won't be
until this weekend that he will have a chance to test, and I'm 
guessing the DT API patch will make its way through Rob's tree to
linux-next by then.

FYI, the warning workaround would be to make the second parameter to
is31fl32xx_parse_child_dt() non-const. 

[1] https://lkml.org/lkml/2016/3/3/924

> > Changes from RFC:
> >   - Removed max-brightness DT property.
> >   - Refer to these devices as "LED controllers" in Kconfig.
> >   - Removed redundant last sentence from Kconfig entry
> >   - Removed unnecessary debug code.
> >   - Do not set led_classdev.brightness to 0 explicitly, as it is
> >     already initialized to 0 by devm_kzalloc().
> >   - Used of_property_read_string() instead of of_get_property().
> >   - Fail immediately on DT parsing error in a child node, rather than
> >     continuing on with the non-faulty ones.
> >   - Added additional comments for some things that might be non-obvious.
> >   - Added constants for the location of the SSD bit in the SHUTDOWN
> >     register, and the 3216's CONFIG register.
> >   - Added special sw_shutdown_func for the 3216 device, as that bit
> >     is in a different register, at a different position, and has reverse
> >     polarity compared to all the other devices.
> >   - Refactored is31fl32xx_init_regs() to separate out some logic into
> >     is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
> >
> > [1] https://lkml.org/lkml/2016/3/2/746
> >
> >   drivers/leds/Kconfig           |   8 +
> >   drivers/leds/Makefile          |   1 +
> >   drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
> >   3 files changed, 514 insertions(+)
> >   create mode 100644 drivers/leds/leds-is31fl32xx.c
> >
> > diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> > index 1034696..9c63ba4 100644
> > --- a/drivers/leds/Kconfig
> > +++ b/drivers/leds/Kconfig
> > @@ -580,6 +580,14 @@ config LEDS_SN3218
> >   	  This driver can also be built as a module. If so the module
> >   	  will be called leds-sn3218.
> >
> > +config LEDS_IS31FL32XX
> > +	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
> > +	depends on LEDS_CLASS && I2C && OF
> > +	help
> > +	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
> > +	  They are I2C devices with multiple constant-current channels, each
> > +	  with independent 256-level PWM control.
> > +
> >   comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
> >
> >   config LEDS_BLINKM
> > diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
> > index 89c9b6f..3fdf313 100644
> > --- a/drivers/leds/Makefile
> > +++ b/drivers/leds/Makefile
> > @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
> >   obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
> >   obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
> >   obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
> > +obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
> >
> >   # LED SPI Drivers
> >   obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
> > diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
> > new file mode 100644
> > index 0000000..49818f0
> > --- /dev/null
> > +++ b/drivers/leds/leds-is31fl32xx.c
> > @@ -0,0 +1,505 @@
> > +/*
> > + * linux/drivers/leds-is31fl32xx.c
> > + *
> > + * Driver for ISSI IS31FL32xx family of I2C LED controllers
> > + *
> > + * Copyright 2015 Allworx Corp.
> > + *
> > + *
> > + * 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.
> > + *
> > + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> > + */
> > +
> > +#include <linux/err.h>
> > +#include <linux/i2c.h>
> > +#include <linux/kernel.h>
> > +#include <linux/leds.h>
> > +#include <linux/module.h>
> > +#include <linux/of_platform.h>
> > +
> > +/* Used to indicate a device has no such register */
> > +#define IS31FL32XX_REG_NONE 0xFF
> > +
> > +/* Software Shutdown bit in Shutdown Register */
> > +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
> > +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
> > +
> > +/* IS31FL3216 has a number of unique registers */
> > +#define IS31FL3216_CONFIG_REG 0x00
> > +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
> > +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
> > +
> > +/* Software Shutdown bit in 3216 Config Register */
> > +#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
> > +#define IS31FL3216_CONFIG_SSD_DISABLE 0
> > +
> > +struct is31fl32xx_priv;
> > +struct is31fl32xx_led_data {
> > +	struct led_classdev cdev;
> > +	u8 channel; /* 1-based, max priv->cdef->channels */
> > +	struct is31fl32xx_priv *priv;
> > +};
> > +
> > +struct is31fl32xx_priv {
> > +	const struct is31fl32xx_chipdef *cdef;
> > +	struct i2c_client *client;
> > +	unsigned int num_leds;
> > +	struct is31fl32xx_led_data leds[0];
> > +};
> > +
> > +/**
> > + * struct is31fl32xx_chipdef - chip-specific attributes
> > + * @channels            : Number of LED channels
> > + * @shutdown_reg        : address of Shutdown register (optional)
> > + * @pwm_update_reg      : address of PWM Update register
> > + * @global_control_reg  : address of Global Control register (optional)
> > + * @reset_reg           : address of Reset register (optional)
> > + * @pwm_register_base   : address of first PWM register
> > + * @pwm_registers_reversed: : true if PWM registers count down instead of up
> > + * @led_control_register_base : address of first LED control register (optional)
> > + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
> > + * @reset_func:         : pointer to reset function
> > + *
> > + * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
> > + * indicates that this chip has no such register.
> > + *
> > + * If non-NULL, @reset_func will be called during probing to set all
> > + * necessary registers to a known initialization state. This is needed
> > + * for chips that do not have a @reset_reg.
> > + *
> > + * @enable_bits_per_led_control_register must be >=1 if
> > + * @led_control_register_base != %IS31FL32XX_REG_NONE.
> > + */
> > +struct is31fl32xx_chipdef {
> > +	u8	channels;
> > +	u8	shutdown_reg;
> > +	u8	pwm_update_reg;
> > +	u8	global_control_reg;
> > +	u8	reset_reg;
> > +	u8	pwm_register_base;
> > +	bool	pwm_registers_reversed;
> > +	u8	led_control_register_base;
> > +	u8	enable_bits_per_led_control_register;
> > +	int (*reset_func)(struct is31fl32xx_priv *priv);
> > +	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
> > +};
> > +
> > +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
> > +	.channels				= 36,
> > +	.shutdown_reg				= 0x00,
> > +	.pwm_update_reg				= 0x25,
> > +	.global_control_reg			= 0x4a,
> > +	.reset_reg				= 0x4f,
> > +	.pwm_register_base			= 0x01,
> > +	.led_control_register_base		= 0x26,
> > +	.enable_bits_per_led_control_register	= 1,
> > +};
> > +
> > +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
> > +	.channels				= 28,
> > +	.shutdown_reg				= 0x00,
> > +	.pwm_update_reg				= 0x25,
> > +	.global_control_reg			= 0x4a,
> > +	.reset_reg				= 0x4f,
> > +	.pwm_register_base			= 0x05,
> > +	.led_control_register_base		= 0x2a,
> > +	.enable_bits_per_led_control_register	= 1,
> > +};
> > +
> > +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
> > +	.channels				= 18,
> > +	.shutdown_reg				= 0x00,
> > +	.pwm_update_reg				= 0x16,
> > +	.global_control_reg			= IS31FL32XX_REG_NONE,
> > +	.reset_reg				= 0x17,
> > +	.pwm_register_base			= 0x01,
> > +	.led_control_register_base		= 0x13,
> > +	.enable_bits_per_led_control_register	= 6,
> > +};
> > +
> > +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
> > +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> > +					bool enable);
> > +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
> > +	.channels				= 16,
> > +	.shutdown_reg				= IS31FL32XX_REG_NONE,
> > +	.pwm_update_reg				= 0xB0,
> > +	.global_control_reg			= IS31FL32XX_REG_NONE,
> > +	.reset_reg				= IS31FL32XX_REG_NONE,
> > +	.pwm_register_base			= 0x10,
> > +	.pwm_registers_reversed			= true,
> > +	.led_control_register_base		= 0x01,
> > +	.enable_bits_per_led_control_register	= 8,
> > +	.reset_func				= is31fl3216_reset,
> > +	.sw_shutdown_func			= is31fl3216_software_shutdown,
> > +};
> > +
> > +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
> > +{
> > +	int ret;
> > +
> > +	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
> > +
> > +	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
> > +	if (ret) {
> > +		dev_err(&priv->client->dev,
> > +			"register write to 0x%02X failed (error %d)",
> > +			reg, ret);
> > +	}
> > +	return ret;
> > +}
> > +
> > +/*
> > + * Custom reset function for IS31FL3216 because it does not have a RESET
> > + * register the way that the other IS31FL32xx chips do. We don't bother
> > + * writing the GPIO and animation registers, because the registers we
> > + * do write ensure those will have no effect.
> > + */
> > +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
> > +{
> > +	unsigned int i;
> > +	int ret;
> > +
> > +	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
> > +			       IS31FL3216_CONFIG_SSD_ENABLE);
> > +	if (ret)
> > +		return ret;
> > +	for (i = 0; i < priv->cdef->channels; i++) {
> > +		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
> > +				       0x00);
> > +		if (ret)
> > +			return ret;
> > +	}
> > +	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
> > +	if (ret)
> > +		return ret;
> > +	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
> > +	if (ret)
> > +		return ret;
> > +	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
> > +	if (ret)
> > +		return ret;
> > +
> > +	return 0;
> > +}
> > +
> > +/*
> > + * Custom Software-Shutdown function for IS31FL3216 because it does not have
> > + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
> > + * We don't bother doing a read/modify/write on the CONFIG register because
> > + * we only ever use a value of '0' for the other fields in that register.
> > + */
> > +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> > +					bool enable)
> > +{
> > +	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
> > +			    IS31FL3216_CONFIG_SSD_DISABLE;
> > +
> > +	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
> > +}
> > +
> > +/*
> > + * NOTE: A mutex is not needed in this function because:
> > + * - All referenced data is read-only after probe()
> > + * - The I2C core has a mutex on to protect the bus
> > + * - There are no read/modify/write operations
> > + * - Intervening operations between the write of the PWM register
> > + *   and the Update register are harmless.
> > + *
> > + * Example:
> > + *	PWM_REG_1 write 16
> > + *	UPDATE_REG write 0
> > + *	PWM_REG_2 write 128
> > + *	UPDATE_REG write 0
> > + *   vs:
> > + *	PWM_REG_1 write 16
> > + *	PWM_REG_2 write 128
> > + *	UPDATE_REG write 0
> > + *	UPDATE_REG write 0
> > + * are equivalent. Poking the Update register merely applies all PWM
> > + * register writes up to that point.
> > + */
> > +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
> > +				     enum led_brightness brightness)
> > +{
> > +	const struct is31fl32xx_led_data *led_data =
> > +		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
> > +	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
> > +	u8 pwm_register_offset;
> > +	int ret;
> > +
> > +	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
> > +
> > +	/* NOTE: led_data->channel is 1-based */
> > +	if (cdef->pwm_registers_reversed)
> > +		pwm_register_offset = cdef->channels - led_data->channel;
> > +	else
> > +		pwm_register_offset = led_data->channel - 1;
> > +
> > +	ret = is31fl32xx_write(led_data->priv,
> > +			       cdef->pwm_register_base + pwm_register_offset,
> > +			       brightness);
> > +	if (ret)
> > +		return ret;
> > +
> > +	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
> > +}
> > +
> > +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
> > +{
> > +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> > +	int ret;
> > +
> > +	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
> > +		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
> > +		if (ret)
> > +			return ret;
> > +	}
> > +
> > +	if (cdef->reset_func)
> > +		return cdef->reset_func(priv);
> > +
> > +	return 0;
> > +}
> > +
> > +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
> > +					bool enable)
> > +{
> > +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> > +	int ret;
> > +
> > +	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
> > +		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
> > +				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
> > +		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
> > +		if (ret)
> > +			return ret;
> > +	}
> > +
> > +	if (cdef->sw_shutdown_func)
> > +		return cdef->sw_shutdown_func(priv, enable);  
> 
> You seem to call sw_shutdown_func only here, so why should we have
> enable parameter in this op?

I'm not sure if I understand the question, but I will try to answer.

'enable' is passed through is31fl32xx_software_shutdown to 
cdef->sw_shutdown_func, so it can be either true or false at that 
point. The purpose of sw_shutdown_func is to add any special behavior 
when enabling/disabling software-shutdown mode, which is needed for 
the 3216 because its SSD bit is in a different position and with 
opposite polarity.

Is it that 'enable' in that line of code makes it look like it's being 
called with an hardcoded value rather than a variable? If so, perhaps a 
different parameter name would make it more obvious? Or a kerneldoc
comment for the function to describe the parameter?

Or have I totally missed the point of the question?

> > +	return 0;
> > +}
> > +
> > +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
> > +{
> > +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> > +	int ret;
> > +
> > +	ret = is31fl32xx_reset_regs(priv);
> > +	if (ret)
> > +		return ret;
> > +
> > +	/*
> > +	 * Set enable bit for all channels.
> > +	 * We will control state with PWM registers alone.
> > +	 */
> > +	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
> > +		u8 value =
> > +		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
> > +		u8 num_regs = cdef->channels /
> > +				cdef->enable_bits_per_led_control_register;
> > +		int i;
> > +
> > +		for (i = 0; i < num_regs; i++) {
> > +			ret = is31fl32xx_write(priv,
> > +					       cdef->led_control_register_base+i,
> > +					       value);
> > +			if (ret)
> > +				return ret;
> > +		}
> > +	}
> > +
> > +	ret = is31fl32xx_software_shutdown(priv, false);
> > +	if (ret)
> > +		return ret;
> > +
> > +	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
> > +		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
> > +		if (ret)
> > +			return ret;
> > +	}
> > +
> > +	return 0;
> > +}
> > +
> > +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
> > +{
> > +	return sizeof(struct is31fl32xx_priv) +
> > +		      (sizeof(struct is31fl32xx_led_data) * num_leds);
> > +}
> > +
> > +static int is31fl32xx_parse_child_dt(const struct device *dev,
> > +				     const struct device_node *child,
> > +				     struct is31fl32xx_led_data *led_data)
> > +{
> > +	struct led_classdev *cdev = &led_data->cdev;
> > +	int ret = 0;
> > +	u32 reg;
> > +
> > +	if (of_property_read_string(child, "label", &cdev->name))
> > +		cdev->name = child->name;
> > +
> > +	ret = of_property_read_u32(child, "reg", &reg);
> > +	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
> > +		dev_err(dev,
> > +			"Child node %s does not have a valid reg property\n",
> > +			child->full_name);
> > +		return -EINVAL;
> > +	}
> > +	led_data->channel = reg;
> > +
> > +	of_property_read_string(child, "linux,default-trigger",
> > +				&cdev->default_trigger);
> > +
> > +	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
> > +
> > +	return 0;
> > +}
> > +
> > +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
> > +					struct is31fl32xx_priv *priv,
> > +					u8 channel)
> > +{
> > +	size_t i;
> > +
> > +	for (i = 0; i < priv->num_leds; i++) {
> > +		if (priv->leds[i].channel == channel)
> > +			return &priv->leds[i];
> > +	}
> > +
> > +	return NULL;
> > +}
> > +
> > +static int is31fl32xx_parse_dt(struct device *dev,
> > +			       struct is31fl32xx_priv *priv)
> > +{
> > +	struct device_node *child;
> > +	int ret = 0;
> > +
> > +	for_each_child_of_node(dev->of_node, child) {
> > +		struct is31fl32xx_led_data *led_data =
> > +			&priv->leds[priv->num_leds];
> > +		const struct is31fl32xx_led_data *other_led_data;
> > +
> > +		led_data->priv = priv;
> > +
> > +		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
> > +		if (ret)
> > +			goto err;
> > +
> > +		/* Detect if channel is already in use by another child */
> > +		other_led_data = is31fl32xx_find_led_data(priv,
> > +							  led_data->channel);
> > +		if (other_led_data) {
> > +			dev_err(dev,
> > +				"%s and %s both attempting to use channel %d\n",
> > +				led_data->cdev.name,
> > +				other_led_data->cdev.name,
> > +				led_data->channel);
> > +			goto err;
> > +		}
> > +
> > +		ret = devm_led_classdev_register(dev, &led_data->cdev);
> > +		if (ret) {
> > +			dev_err(dev, "failed to register PWM led for %s: %d\n",
> > +				led_data->cdev.name, ret);
> > +			goto err;
> > +		}
> > +
> > +		priv->num_leds++;
> > +	}
> > +
> > +	return 0;
> > +
> > +err:
> > +	of_node_put(child);
> > +	return ret;
> > +}
> > +
> > +static const struct of_device_id of_is31fl31xx_match[] = {
> > +	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
> > +	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
> > +	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
> > +	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
> > +	{},
> > +};
> > +
> > +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
> > +
> > +static int is31fl32xx_probe(struct i2c_client *client,
> > +			    const struct i2c_device_id *id)
> > +{
> > +	const struct is31fl32xx_chipdef *cdef;
> > +	const struct of_device_id *of_dev_id;
> > +	struct device *dev = &client->dev;
> > +	struct is31fl32xx_priv *priv;
> > +	int count;
> > +	int ret = 0;
> > +
> > +	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
> > +	if (!of_dev_id)
> > +		return -EINVAL;
> > +
> > +	cdef = of_dev_id->data;
> > +
> > +	count = of_get_child_count(dev->of_node);
> > +	if (!count)
> > +		return -EINVAL;
> > +
> > +	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
> > +			    GFP_KERNEL);
> > +	if (!priv)
> > +		return -ENOMEM;
> > +
> > +	priv->client = client;
> > +	priv->cdef = cdef;
> > +	i2c_set_clientdata(client, priv);
> > +
> > +	ret = is31fl32xx_init_regs(priv);
> > +	if (ret)
> > +		return ret;
> > +
> > +	ret = is31fl32xx_parse_dt(dev, priv);
> > +	if (ret)
> > +		return ret;
> > +
> > +	return 0;
> > +}
> > +
> > +static int is31fl32xx_remove(struct i2c_client *client)
> > +{
> > +	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
> > +
> > +	return is31fl32xx_reset_regs(priv);
> > +}
> > +
> > +/*
> > + * i2c-core requires that id_table be non-NULL, even though
> > + * it is not used for DeviceTree based instantiation.
> > + */
> > +static const struct i2c_device_id is31fl31xx_id[] = {
> > +	{},
> > +};
> > +
> > +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
> > +
> > +static struct i2c_driver is31fl32xx_driver = {
> > +	.driver = {
> > +		.name	= "is31fl32xx",
> > +		.of_match_table = of_is31fl31xx_match,
> > +	},
> > +	.probe		= is31fl32xx_probe,
> > +	.remove		= is31fl32xx_remove,
> > +	.id_table	= is31fl31xx_id,
> > +};
> > +
> > +module_i2c_driver(is31fl32xx_driver);
> > +
> > +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
> > +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
> > +MODULE_LICENSE("GPL v2");
> >  

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[Jacek Anaszewski]

On 03/04/2016 01:45 AM, David Rivshin (Allworx) wrote:
> On Thu, 03 Mar 2016 15:51:32 +0100
> Jacek Anaszewski <j.anaszewski-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org> wrote:
>
>> Hi David,
>>
>> Thanks for the update. Two remarks in the code.
>>
>> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:
>>> From: David Rivshin <drivshin-5fOYsn7Fw8lBDgjK7y7TUQ@public.gmane.org>
>>>
>>> The IS31FL32xx family of LED controllers are I2C devices with multiple
>>> constant-current channels, each with independent 256-level PWM control.
>>>
>>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>>
>>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
>>> (TI am335x) platform.
>>>
>>> The programming paradigm of these devices is similar in the following
>>> ways:
>>>    - All registers are 8 bit
>>>    - All LED control registers are write-only
>>>    - Each LED channel has a PWM register (0-255)
>>>    - PWM register writes are shadowed until an Update register is poked
>>>    - All have a concept of Software Shutdown, which disables output
>>>
>>> However, there are some differences in devices:
>>>    - 3236/3235 have a separate Control register for each LED,
>>>      (3218/3216 pack the enable bits into fewer registers)
>>>    - 3236/3235 have a per-channel current divisor setting
>>>    - 3236/3235 have a Global Control register that can turn off all LEDs
>>>    - 3216 is unique in a number of ways
>>>       - OUT9-OUT16 can be configured as GPIOs instead of LED controls
>>>       - LEDs can be programmed with an 8-frame animation, with
>>>         programmable delay between frames
>>>       - LEDs can be modulated by an input audio signal
>>>       - Max output current can be adjusted from 1/4 to 2x globally
>>>       - Has a Configuration register instead of a Shutdown register
>>>
>>> This driver currently only supports the base PWM control function
>>> of these devices. The following features of these devices are not
>>> implemented, although it should be possible to add them in the future:
>>>    - All devices are capable of going into a lower-power "software
>>>      shutdown" mode.
>>>    - The is31fl3236 and is31fl3235 can reduce the max output current
>>>      per-channel with a divisor of 1, 2, 3, or 4.
>>>    - The is31fl3216 can use some LED channels as GPIOs instead.
>>>    - The is31fl3216 can animate LEDs in hardware.
>>>    - The is31fl3216 can modulate LEDs according to an audio input.
>>>    - The is31fl3216 can reduce/increase max output current globally.
>>>
>>> Signed-off-by: David Rivshin <drivshin-5fOYsn7Fw8lBDgjK7y7TUQ@public.gmane.org>
>>> ---
>>>
>>> You may see two instances of this warning:
>>>     "passing argument 1 of 'of_property_read_string' discards 'const'
>>>      qualifier from pointer target type"
>>> That is a result of of_property_read_string() taking a non-const
>>> struct device_node pointer parameter. I have separately submitted a
>>> patch to fix that [1], and a few related functions which had the same
>>> issue. I'm hoping that will get into linux-next before this does, so
>>> that the warnings never show up there.
>>
>> Please adjust the patch so that it compiles without warnings on
>> current linux-next. Your patch for DT API hasn't been reviewed yet
>> AFICS, and I can imagine that there will be some resistance against.
>
> Since the DT API patch was just accepted by Rob [1], would it be OK
> to wait for the results of Stefan's testing (and any other reviews)
> before making a decision on this? From Stefan's note, it won't be
> until this weekend that he will have a chance to test, and I'm
> guessing the DT API patch will make its way through Rob's tree to
> linux-next by then.

OK.

> FYI, the warning workaround would be to make the second parameter to
> is31fl32xx_parse_child_dt() non-const.
>
> [1] https://lkml.org/lkml/2016/3/3/924
>
>>> Changes from RFC:
>>>    - Removed max-brightness DT property.
>>>    - Refer to these devices as "LED controllers" in Kconfig.
>>>    - Removed redundant last sentence from Kconfig entry
>>>    - Removed unnecessary debug code.
>>>    - Do not set led_classdev.brightness to 0 explicitly, as it is
>>>      already initialized to 0 by devm_kzalloc().
>>>    - Used of_property_read_string() instead of of_get_property().
>>>    - Fail immediately on DT parsing error in a child node, rather than
>>>      continuing on with the non-faulty ones.
>>>    - Added additional comments for some things that might be non-obvious.
>>>    - Added constants for the location of the SSD bit in the SHUTDOWN
>>>      register, and the 3216's CONFIG register.
>>>    - Added special sw_shutdown_func for the 3216 device, as that bit
>>>      is in a different register, at a different position, and has reverse
>>>      polarity compared to all the other devices.
>>>    - Refactored is31fl32xx_init_regs() to separate out some logic into
>>>      is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
>>>
>>> [1] https://lkml.org/lkml/2016/3/2/746
>>>
>>>    drivers/leds/Kconfig           |   8 +
>>>    drivers/leds/Makefile          |   1 +
>>>    drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
>>>    3 files changed, 514 insertions(+)
>>>    create mode 100644 drivers/leds/leds-is31fl32xx.c
>>>
>>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
>>> index 1034696..9c63ba4 100644
>>> --- a/drivers/leds/Kconfig
>>> +++ b/drivers/leds/Kconfig
>>> @@ -580,6 +580,14 @@ config LEDS_SN3218
>>>    	  This driver can also be built as a module. If so the module
>>>    	  will be called leds-sn3218.
>>>
>>> +config LEDS_IS31FL32XX
>>> +	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
>>> +	depends on LEDS_CLASS && I2C && OF
>>> +	help
>>> +	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
>>> +	  They are I2C devices with multiple constant-current channels, each
>>> +	  with independent 256-level PWM control.
>>> +
>>>    comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
>>>
>>>    config LEDS_BLINKM
>>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
>>> index 89c9b6f..3fdf313 100644
>>> --- a/drivers/leds/Makefile
>>> +++ b/drivers/leds/Makefile
>>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
>>>    obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
>>>    obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
>>>    obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
>>> +obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
>>>
>>>    # LED SPI Drivers
>>>    obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
>>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
>>> new file mode 100644
>>> index 0000000..49818f0
>>> --- /dev/null
>>> +++ b/drivers/leds/leds-is31fl32xx.c
>>> @@ -0,0 +1,505 @@
>>> +/*
>>> + * linux/drivers/leds-is31fl32xx.c
>>> + *
>>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
>>> + *
>>> + * Copyright 2015 Allworx Corp.
>>> + *
>>> + *
>>> + * 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.
>>> + *
>>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>> + */
>>> +
>>> +#include <linux/err.h>
>>> +#include <linux/i2c.h>
>>> +#include <linux/kernel.h>
>>> +#include <linux/leds.h>
>>> +#include <linux/module.h>
>>> +#include <linux/of_platform.h>
>>> +
>>> +/* Used to indicate a device has no such register */
>>> +#define IS31FL32XX_REG_NONE 0xFF
>>> +
>>> +/* Software Shutdown bit in Shutdown Register */
>>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
>>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
>>> +
>>> +/* IS31FL3216 has a number of unique registers */
>>> +#define IS31FL3216_CONFIG_REG 0x00
>>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
>>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
>>> +
>>> +/* Software Shutdown bit in 3216 Config Register */
>>> +#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
>>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
>>> +
>>> +struct is31fl32xx_priv;
>>> +struct is31fl32xx_led_data {
>>> +	struct led_classdev cdev;
>>> +	u8 channel; /* 1-based, max priv->cdef->channels */
>>> +	struct is31fl32xx_priv *priv;
>>> +};
>>> +
>>> +struct is31fl32xx_priv {
>>> +	const struct is31fl32xx_chipdef *cdef;
>>> +	struct i2c_client *client;
>>> +	unsigned int num_leds;
>>> +	struct is31fl32xx_led_data leds[0];
>>> +};
>>> +
>>> +/**
>>> + * struct is31fl32xx_chipdef - chip-specific attributes
>>> + * @channels            : Number of LED channels
>>> + * @shutdown_reg        : address of Shutdown register (optional)
>>> + * @pwm_update_reg      : address of PWM Update register
>>> + * @global_control_reg  : address of Global Control register (optional)
>>> + * @reset_reg           : address of Reset register (optional)
>>> + * @pwm_register_base   : address of first PWM register
>>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
>>> + * @led_control_register_base : address of first LED control register (optional)
>>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
>>> + * @reset_func:         : pointer to reset function
>>> + *
>>> + * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
>>> + * indicates that this chip has no such register.
>>> + *
>>> + * If non-NULL, @reset_func will be called during probing to set all
>>> + * necessary registers to a known initialization state. This is needed
>>> + * for chips that do not have a @reset_reg.
>>> + *
>>> + * @enable_bits_per_led_control_register must be >=1 if
>>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
>>> + */
>>> +struct is31fl32xx_chipdef {
>>> +	u8	channels;
>>> +	u8	shutdown_reg;
>>> +	u8	pwm_update_reg;
>>> +	u8	global_control_reg;
>>> +	u8	reset_reg;
>>> +	u8	pwm_register_base;
>>> +	bool	pwm_registers_reversed;
>>> +	u8	led_control_register_base;
>>> +	u8	enable_bits_per_led_control_register;
>>> +	int (*reset_func)(struct is31fl32xx_priv *priv);
>>> +	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
>>> +};
>>> +
>>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
>>> +	.channels				= 36,
>>> +	.shutdown_reg				= 0x00,
>>> +	.pwm_update_reg				= 0x25,
>>> +	.global_control_reg			= 0x4a,
>>> +	.reset_reg				= 0x4f,
>>> +	.pwm_register_base			= 0x01,
>>> +	.led_control_register_base		= 0x26,
>>> +	.enable_bits_per_led_control_register	= 1,
>>> +};
>>> +
>>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
>>> +	.channels				= 28,
>>> +	.shutdown_reg				= 0x00,
>>> +	.pwm_update_reg				= 0x25,
>>> +	.global_control_reg			= 0x4a,
>>> +	.reset_reg				= 0x4f,
>>> +	.pwm_register_base			= 0x05,
>>> +	.led_control_register_base		= 0x2a,
>>> +	.enable_bits_per_led_control_register	= 1,
>>> +};
>>> +
>>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
>>> +	.channels				= 18,
>>> +	.shutdown_reg				= 0x00,
>>> +	.pwm_update_reg				= 0x16,
>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
>>> +	.reset_reg				= 0x17,
>>> +	.pwm_register_base			= 0x01,
>>> +	.led_control_register_base		= 0x13,
>>> +	.enable_bits_per_led_control_register	= 6,
>>> +};
>>> +
>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>> +					bool enable);
>>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
>>> +	.channels				= 16,
>>> +	.shutdown_reg				= IS31FL32XX_REG_NONE,
>>> +	.pwm_update_reg				= 0xB0,
>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
>>> +	.reset_reg				= IS31FL32XX_REG_NONE,
>>> +	.pwm_register_base			= 0x10,
>>> +	.pwm_registers_reversed			= true,
>>> +	.led_control_register_base		= 0x01,
>>> +	.enable_bits_per_led_control_register	= 8,
>>> +	.reset_func				= is31fl3216_reset,
>>> +	.sw_shutdown_func			= is31fl3216_software_shutdown,
>>> +};
>>> +
>>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
>>> +{
>>> +	int ret;
>>> +
>>> +	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
>>> +
>>> +	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
>>> +	if (ret) {
>>> +		dev_err(&priv->client->dev,
>>> +			"register write to 0x%02X failed (error %d)",
>>> +			reg, ret);
>>> +	}
>>> +	return ret;
>>> +}
>>> +
>>> +/*
>>> + * Custom reset function for IS31FL3216 because it does not have a RESET
>>> + * register the way that the other IS31FL32xx chips do. We don't bother
>>> + * writing the GPIO and animation registers, because the registers we
>>> + * do write ensure those will have no effect.
>>> + */
>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
>>> +{
>>> +	unsigned int i;
>>> +	int ret;
>>> +
>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
>>> +			       IS31FL3216_CONFIG_SSD_ENABLE);
>>> +	if (ret)
>>> +		return ret;
>>> +	for (i = 0; i < priv->cdef->channels; i++) {
>>> +		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
>>> +				       0x00);
>>> +		if (ret)
>>> +			return ret;
>>> +	}
>>> +	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
>>> +	if (ret)
>>> +		return ret;
>>> +	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
>>> +	if (ret)
>>> +		return ret;
>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +/*
>>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
>>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
>>> + * We don't bother doing a read/modify/write on the CONFIG register because
>>> + * we only ever use a value of '0' for the other fields in that register.
>>> + */
>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>> +					bool enable)
>>> +{
>>> +	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
>>> +			    IS31FL3216_CONFIG_SSD_DISABLE;
>>> +
>>> +	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
>>> +}
>>> +
>>> +/*
>>> + * NOTE: A mutex is not needed in this function because:
>>> + * - All referenced data is read-only after probe()
>>> + * - The I2C core has a mutex on to protect the bus
>>> + * - There are no read/modify/write operations
>>> + * - Intervening operations between the write of the PWM register
>>> + *   and the Update register are harmless.
>>> + *
>>> + * Example:
>>> + *	PWM_REG_1 write 16
>>> + *	UPDATE_REG write 0
>>> + *	PWM_REG_2 write 128
>>> + *	UPDATE_REG write 0
>>> + *   vs:
>>> + *	PWM_REG_1 write 16
>>> + *	PWM_REG_2 write 128
>>> + *	UPDATE_REG write 0
>>> + *	UPDATE_REG write 0
>>> + * are equivalent. Poking the Update register merely applies all PWM
>>> + * register writes up to that point.
>>> + */
>>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
>>> +				     enum led_brightness brightness)
>>> +{
>>> +	const struct is31fl32xx_led_data *led_data =
>>> +		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
>>> +	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
>>> +	u8 pwm_register_offset;
>>> +	int ret;
>>> +
>>> +	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
>>> +
>>> +	/* NOTE: led_data->channel is 1-based */
>>> +	if (cdef->pwm_registers_reversed)
>>> +		pwm_register_offset = cdef->channels - led_data->channel;
>>> +	else
>>> +		pwm_register_offset = led_data->channel - 1;
>>> +
>>> +	ret = is31fl32xx_write(led_data->priv,
>>> +			       cdef->pwm_register_base + pwm_register_offset,
>>> +			       brightness);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
>>> +}
>>> +
>>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
>>> +{
>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>> +	int ret;
>>> +
>>> +	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
>>> +		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
>>> +		if (ret)
>>> +			return ret;
>>> +	}
>>> +
>>> +	if (cdef->reset_func)
>>> +		return cdef->reset_func(priv);
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
>>> +					bool enable)
>>> +{
>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>> +	int ret;
>>> +
>>> +	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
>>> +		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
>>> +				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
>>> +		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
>>> +		if (ret)
>>> +			return ret;
>>> +	}
>>> +
>>> +	if (cdef->sw_shutdown_func)
>>> +		return cdef->sw_shutdown_func(priv, enable);
>>
>> You seem to call sw_shutdown_func only here, so why should we have
>> enable parameter in this op?
>
> I'm not sure if I understand the question, but I will try to answer.
>
> 'enable' is passed through is31fl32xx_software_shutdown to
> cdef->sw_shutdown_func, so it can be either true or false at that
> point. The purpose of sw_shutdown_func is to add any special behavior
> when enabling/disabling software-shutdown mode, which is needed for
> the 3216 because its SSD bit is in a different position and with
> opposite polarity.
>
> Is it that 'enable' in that line of code makes it look like it's being
> called with an hardcoded value rather than a variable? If so, perhaps a
> different parameter name would make it more obvious? Or a kerneldoc
> comment for the function to describe the parameter?
>
> Or have I totally missed the point of the question?

Actually I should have placed this question next to
the call to s31fl32xx_software_shutdown() in is31fl32xx_init_regs(),
which is passed "false" in the second argument, and there is no
other call to s31fl32xx_software_shutdown() in the driver.

Having the argument makes people wondering that there is some
use case in the driver, where "true" is passed, but it seems not
to be the case.

>>> +	return 0;
>>> +}
>>> +
>>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
>>> +{
>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>> +	int ret;
>>> +
>>> +	ret = is31fl32xx_reset_regs(priv);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	/*
>>> +	 * Set enable bit for all channels.
>>> +	 * We will control state with PWM registers alone.
>>> +	 */
>>> +	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
>>> +		u8 value =
>>> +		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
>>> +		u8 num_regs = cdef->channels /
>>> +				cdef->enable_bits_per_led_control_register;
>>> +		int i;
>>> +
>>> +		for (i = 0; i < num_regs; i++) {
>>> +			ret = is31fl32xx_write(priv,
>>> +					       cdef->led_control_register_base+i,
>>> +					       value);
>>> +			if (ret)
>>> +				return ret;
>>> +		}
>>> +	}
>>> +
>>> +	ret = is31fl32xx_software_shutdown(priv, false);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
>>> +		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
>>> +		if (ret)
>>> +			return ret;
>>> +	}
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
>>> +{
>>> +	return sizeof(struct is31fl32xx_priv) +
>>> +		      (sizeof(struct is31fl32xx_led_data) * num_leds);
>>> +}
>>> +
>>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
>>> +				     const struct device_node *child,
>>> +				     struct is31fl32xx_led_data *led_data)
>>> +{
>>> +	struct led_classdev *cdev = &led_data->cdev;
>>> +	int ret = 0;
>>> +	u32 reg;
>>> +
>>> +	if (of_property_read_string(child, "label", &cdev->name))
>>> +		cdev->name = child->name;
>>> +
>>> +	ret = of_property_read_u32(child, "reg", &reg);
>>> +	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
>>> +		dev_err(dev,
>>> +			"Child node %s does not have a valid reg property\n",
>>> +			child->full_name);
>>> +		return -EINVAL;
>>> +	}
>>> +	led_data->channel = reg;
>>> +
>>> +	of_property_read_string(child, "linux,default-trigger",
>>> +				&cdev->default_trigger);
>>> +
>>> +	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
>>> +					struct is31fl32xx_priv *priv,
>>> +					u8 channel)
>>> +{
>>> +	size_t i;
>>> +
>>> +	for (i = 0; i < priv->num_leds; i++) {
>>> +		if (priv->leds[i].channel == channel)
>>> +			return &priv->leds[i];
>>> +	}
>>> +
>>> +	return NULL;
>>> +}
>>> +
>>> +static int is31fl32xx_parse_dt(struct device *dev,
>>> +			       struct is31fl32xx_priv *priv)
>>> +{
>>> +	struct device_node *child;
>>> +	int ret = 0;
>>> +
>>> +	for_each_child_of_node(dev->of_node, child) {
>>> +		struct is31fl32xx_led_data *led_data =
>>> +			&priv->leds[priv->num_leds];
>>> +		const struct is31fl32xx_led_data *other_led_data;
>>> +
>>> +		led_data->priv = priv;
>>> +
>>> +		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
>>> +		if (ret)
>>> +			goto err;
>>> +
>>> +		/* Detect if channel is already in use by another child */
>>> +		other_led_data = is31fl32xx_find_led_data(priv,
>>> +							  led_data->channel);
>>> +		if (other_led_data) {
>>> +			dev_err(dev,
>>> +				"%s and %s both attempting to use channel %d\n",
>>> +				led_data->cdev.name,
>>> +				other_led_data->cdev.name,
>>> +				led_data->channel);
>>> +			goto err;
>>> +		}
>>> +
>>> +		ret = devm_led_classdev_register(dev, &led_data->cdev);
>>> +		if (ret) {
>>> +			dev_err(dev, "failed to register PWM led for %s: %d\n",
>>> +				led_data->cdev.name, ret);
>>> +			goto err;
>>> +		}
>>> +
>>> +		priv->num_leds++;
>>> +	}
>>> +
>>> +	return 0;
>>> +
>>> +err:
>>> +	of_node_put(child);
>>> +	return ret;
>>> +}
>>> +
>>> +static const struct of_device_id of_is31fl31xx_match[] = {
>>> +	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
>>> +	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
>>> +	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
>>> +	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
>>> +	{},
>>> +};
>>> +
>>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
>>> +
>>> +static int is31fl32xx_probe(struct i2c_client *client,
>>> +			    const struct i2c_device_id *id)
>>> +{
>>> +	const struct is31fl32xx_chipdef *cdef;
>>> +	const struct of_device_id *of_dev_id;
>>> +	struct device *dev = &client->dev;
>>> +	struct is31fl32xx_priv *priv;
>>> +	int count;
>>> +	int ret = 0;
>>> +
>>> +	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
>>> +	if (!of_dev_id)
>>> +		return -EINVAL;
>>> +
>>> +	cdef = of_dev_id->data;
>>> +
>>> +	count = of_get_child_count(dev->of_node);
>>> +	if (!count)
>>> +		return -EINVAL;
>>> +
>>> +	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
>>> +			    GFP_KERNEL);
>>> +	if (!priv)
>>> +		return -ENOMEM;
>>> +
>>> +	priv->client = client;
>>> +	priv->cdef = cdef;
>>> +	i2c_set_clientdata(client, priv);
>>> +
>>> +	ret = is31fl32xx_init_regs(priv);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	ret = is31fl32xx_parse_dt(dev, priv);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +static int is31fl32xx_remove(struct i2c_client *client)
>>> +{
>>> +	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
>>> +
>>> +	return is31fl32xx_reset_regs(priv);
>>> +}
>>> +
>>> +/*
>>> + * i2c-core requires that id_table be non-NULL, even though
>>> + * it is not used for DeviceTree based instantiation.
>>> + */
>>> +static const struct i2c_device_id is31fl31xx_id[] = {
>>> +	{},
>>> +};
>>> +
>>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
>>> +
>>> +static struct i2c_driver is31fl32xx_driver = {
>>> +	.driver = {
>>> +		.name	= "is31fl32xx",
>>> +		.of_match_table = of_is31fl31xx_match,
>>> +	},
>>> +	.probe		= is31fl32xx_probe,
>>> +	.remove		= is31fl32xx_remove,
>>> +	.id_table	= is31fl31xx_id,
>>> +};
>>> +
>>> +module_i2c_driver(is31fl32xx_driver);
>>> +
>>> +MODULE_AUTHOR("David Rivshin <drivshin-5fOYsn7Fw8lBDgjK7y7TUQ@public.gmane.org>");
>>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
>>> +MODULE_LICENSE("GPL v2");
>>>
>
>
>


-- 
Best regards,
Jacek Anaszewski
--
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 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[Jacek Anaszewski]

On 03/04/2016 01:45 AM, David Rivshin (Allworx) wrote:
> On Thu, 03 Mar 2016 15:51:32 +0100
> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>
>> Hi David,
>>
>> Thanks for the update. Two remarks in the code.
>>
>> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:
>>> From: David Rivshin <drivshin@allworx.com>
>>>
>>> The IS31FL32xx family of LED controllers are I2C devices with multiple
>>> constant-current channels, each with independent 256-level PWM control.
>>>
>>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>>
>>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
>>> (TI am335x) platform.
>>>
>>> The programming paradigm of these devices is similar in the following
>>> ways:
>>>    - All registers are 8 bit
>>>    - All LED control registers are write-only
>>>    - Each LED channel has a PWM register (0-255)
>>>    - PWM register writes are shadowed until an Update register is poked
>>>    - All have a concept of Software Shutdown, which disables output
>>>
>>> However, there are some differences in devices:
>>>    - 3236/3235 have a separate Control register for each LED,
>>>      (3218/3216 pack the enable bits into fewer registers)
>>>    - 3236/3235 have a per-channel current divisor setting
>>>    - 3236/3235 have a Global Control register that can turn off all LEDs
>>>    - 3216 is unique in a number of ways
>>>       - OUT9-OUT16 can be configured as GPIOs instead of LED controls
>>>       - LEDs can be programmed with an 8-frame animation, with
>>>         programmable delay between frames
>>>       - LEDs can be modulated by an input audio signal
>>>       - Max output current can be adjusted from 1/4 to 2x globally
>>>       - Has a Configuration register instead of a Shutdown register
>>>
>>> This driver currently only supports the base PWM control function
>>> of these devices. The following features of these devices are not
>>> implemented, although it should be possible to add them in the future:
>>>    - All devices are capable of going into a lower-power "software
>>>      shutdown" mode.
>>>    - The is31fl3236 and is31fl3235 can reduce the max output current
>>>      per-channel with a divisor of 1, 2, 3, or 4.
>>>    - The is31fl3216 can use some LED channels as GPIOs instead.
>>>    - The is31fl3216 can animate LEDs in hardware.
>>>    - The is31fl3216 can modulate LEDs according to an audio input.
>>>    - The is31fl3216 can reduce/increase max output current globally.
>>>
>>> Signed-off-by: David Rivshin <drivshin@allworx.com>
>>> ---
>>>
>>> You may see two instances of this warning:
>>>     "passing argument 1 of 'of_property_read_string' discards 'const'
>>>      qualifier from pointer target type"
>>> That is a result of of_property_read_string() taking a non-const
>>> struct device_node pointer parameter. I have separately submitted a
>>> patch to fix that [1], and a few related functions which had the same
>>> issue. I'm hoping that will get into linux-next before this does, so
>>> that the warnings never show up there.
>>
>> Please adjust the patch so that it compiles without warnings on
>> current linux-next. Your patch for DT API hasn't been reviewed yet
>> AFICS, and I can imagine that there will be some resistance against.
>
> Since the DT API patch was just accepted by Rob [1], would it be OK
> to wait for the results of Stefan's testing (and any other reviews)
> before making a decision on this? From Stefan's note, it won't be
> until this weekend that he will have a chance to test, and I'm
> guessing the DT API patch will make its way through Rob's tree to
> linux-next by then.

OK.

> FYI, the warning workaround would be to make the second parameter to
> is31fl32xx_parse_child_dt() non-const.
>
> [1] https://lkml.org/lkml/2016/3/3/924
>
>>> Changes from RFC:
>>>    - Removed max-brightness DT property.
>>>    - Refer to these devices as "LED controllers" in Kconfig.
>>>    - Removed redundant last sentence from Kconfig entry
>>>    - Removed unnecessary debug code.
>>>    - Do not set led_classdev.brightness to 0 explicitly, as it is
>>>      already initialized to 0 by devm_kzalloc().
>>>    - Used of_property_read_string() instead of of_get_property().
>>>    - Fail immediately on DT parsing error in a child node, rather than
>>>      continuing on with the non-faulty ones.
>>>    - Added additional comments for some things that might be non-obvious.
>>>    - Added constants for the location of the SSD bit in the SHUTDOWN
>>>      register, and the 3216's CONFIG register.
>>>    - Added special sw_shutdown_func for the 3216 device, as that bit
>>>      is in a different register, at a different position, and has reverse
>>>      polarity compared to all the other devices.
>>>    - Refactored is31fl32xx_init_regs() to separate out some logic into
>>>      is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
>>>
>>> [1] https://lkml.org/lkml/2016/3/2/746
>>>
>>>    drivers/leds/Kconfig           |   8 +
>>>    drivers/leds/Makefile          |   1 +
>>>    drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
>>>    3 files changed, 514 insertions(+)
>>>    create mode 100644 drivers/leds/leds-is31fl32xx.c
>>>
>>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
>>> index 1034696..9c63ba4 100644
>>> --- a/drivers/leds/Kconfig
>>> +++ b/drivers/leds/Kconfig
>>> @@ -580,6 +580,14 @@ config LEDS_SN3218
>>>    	  This driver can also be built as a module. If so the module
>>>    	  will be called leds-sn3218.
>>>
>>> +config LEDS_IS31FL32XX
>>> +	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
>>> +	depends on LEDS_CLASS && I2C && OF
>>> +	help
>>> +	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
>>> +	  They are I2C devices with multiple constant-current channels, each
>>> +	  with independent 256-level PWM control.
>>> +
>>>    comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
>>>
>>>    config LEDS_BLINKM
>>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
>>> index 89c9b6f..3fdf313 100644
>>> --- a/drivers/leds/Makefile
>>> +++ b/drivers/leds/Makefile
>>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
>>>    obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
>>>    obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
>>>    obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
>>> +obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
>>>
>>>    # LED SPI Drivers
>>>    obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
>>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
>>> new file mode 100644
>>> index 0000000..49818f0
>>> --- /dev/null
>>> +++ b/drivers/leds/leds-is31fl32xx.c
>>> @@ -0,0 +1,505 @@
>>> +/*
>>> + * linux/drivers/leds-is31fl32xx.c
>>> + *
>>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
>>> + *
>>> + * Copyright 2015 Allworx Corp.
>>> + *
>>> + *
>>> + * 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.
>>> + *
>>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>> + */
>>> +
>>> +#include <linux/err.h>
>>> +#include <linux/i2c.h>
>>> +#include <linux/kernel.h>
>>> +#include <linux/leds.h>
>>> +#include <linux/module.h>
>>> +#include <linux/of_platform.h>
>>> +
>>> +/* Used to indicate a device has no such register */
>>> +#define IS31FL32XX_REG_NONE 0xFF
>>> +
>>> +/* Software Shutdown bit in Shutdown Register */
>>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
>>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
>>> +
>>> +/* IS31FL3216 has a number of unique registers */
>>> +#define IS31FL3216_CONFIG_REG 0x00
>>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
>>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
>>> +
>>> +/* Software Shutdown bit in 3216 Config Register */
>>> +#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
>>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
>>> +
>>> +struct is31fl32xx_priv;
>>> +struct is31fl32xx_led_data {
>>> +	struct led_classdev cdev;
>>> +	u8 channel; /* 1-based, max priv->cdef->channels */
>>> +	struct is31fl32xx_priv *priv;
>>> +};
>>> +
>>> +struct is31fl32xx_priv {
>>> +	const struct is31fl32xx_chipdef *cdef;
>>> +	struct i2c_client *client;
>>> +	unsigned int num_leds;
>>> +	struct is31fl32xx_led_data leds[0];
>>> +};
>>> +
>>> +/**
>>> + * struct is31fl32xx_chipdef - chip-specific attributes
>>> + * @channels            : Number of LED channels
>>> + * @shutdown_reg        : address of Shutdown register (optional)
>>> + * @pwm_update_reg      : address of PWM Update register
>>> + * @global_control_reg  : address of Global Control register (optional)
>>> + * @reset_reg           : address of Reset register (optional)
>>> + * @pwm_register_base   : address of first PWM register
>>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
>>> + * @led_control_register_base : address of first LED control register (optional)
>>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
>>> + * @reset_func:         : pointer to reset function
>>> + *
>>> + * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
>>> + * indicates that this chip has no such register.
>>> + *
>>> + * If non-NULL, @reset_func will be called during probing to set all
>>> + * necessary registers to a known initialization state. This is needed
>>> + * for chips that do not have a @reset_reg.
>>> + *
>>> + * @enable_bits_per_led_control_register must be >=1 if
>>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
>>> + */
>>> +struct is31fl32xx_chipdef {
>>> +	u8	channels;
>>> +	u8	shutdown_reg;
>>> +	u8	pwm_update_reg;
>>> +	u8	global_control_reg;
>>> +	u8	reset_reg;
>>> +	u8	pwm_register_base;
>>> +	bool	pwm_registers_reversed;
>>> +	u8	led_control_register_base;
>>> +	u8	enable_bits_per_led_control_register;
>>> +	int (*reset_func)(struct is31fl32xx_priv *priv);
>>> +	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
>>> +};
>>> +
>>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
>>> +	.channels				= 36,
>>> +	.shutdown_reg				= 0x00,
>>> +	.pwm_update_reg				= 0x25,
>>> +	.global_control_reg			= 0x4a,
>>> +	.reset_reg				= 0x4f,
>>> +	.pwm_register_base			= 0x01,
>>> +	.led_control_register_base		= 0x26,
>>> +	.enable_bits_per_led_control_register	= 1,
>>> +};
>>> +
>>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
>>> +	.channels				= 28,
>>> +	.shutdown_reg				= 0x00,
>>> +	.pwm_update_reg				= 0x25,
>>> +	.global_control_reg			= 0x4a,
>>> +	.reset_reg				= 0x4f,
>>> +	.pwm_register_base			= 0x05,
>>> +	.led_control_register_base		= 0x2a,
>>> +	.enable_bits_per_led_control_register	= 1,
>>> +};
>>> +
>>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
>>> +	.channels				= 18,
>>> +	.shutdown_reg				= 0x00,
>>> +	.pwm_update_reg				= 0x16,
>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
>>> +	.reset_reg				= 0x17,
>>> +	.pwm_register_base			= 0x01,
>>> +	.led_control_register_base		= 0x13,
>>> +	.enable_bits_per_led_control_register	= 6,
>>> +};
>>> +
>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>> +					bool enable);
>>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
>>> +	.channels				= 16,
>>> +	.shutdown_reg				= IS31FL32XX_REG_NONE,
>>> +	.pwm_update_reg				= 0xB0,
>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
>>> +	.reset_reg				= IS31FL32XX_REG_NONE,
>>> +	.pwm_register_base			= 0x10,
>>> +	.pwm_registers_reversed			= true,
>>> +	.led_control_register_base		= 0x01,
>>> +	.enable_bits_per_led_control_register	= 8,
>>> +	.reset_func				= is31fl3216_reset,
>>> +	.sw_shutdown_func			= is31fl3216_software_shutdown,
>>> +};
>>> +
>>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
>>> +{
>>> +	int ret;
>>> +
>>> +	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
>>> +
>>> +	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
>>> +	if (ret) {
>>> +		dev_err(&priv->client->dev,
>>> +			"register write to 0x%02X failed (error %d)",
>>> +			reg, ret);
>>> +	}
>>> +	return ret;
>>> +}
>>> +
>>> +/*
>>> + * Custom reset function for IS31FL3216 because it does not have a RESET
>>> + * register the way that the other IS31FL32xx chips do. We don't bother
>>> + * writing the GPIO and animation registers, because the registers we
>>> + * do write ensure those will have no effect.
>>> + */
>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
>>> +{
>>> +	unsigned int i;
>>> +	int ret;
>>> +
>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
>>> +			       IS31FL3216_CONFIG_SSD_ENABLE);
>>> +	if (ret)
>>> +		return ret;
>>> +	for (i = 0; i < priv->cdef->channels; i++) {
>>> +		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
>>> +				       0x00);
>>> +		if (ret)
>>> +			return ret;
>>> +	}
>>> +	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
>>> +	if (ret)
>>> +		return ret;
>>> +	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
>>> +	if (ret)
>>> +		return ret;
>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +/*
>>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
>>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
>>> + * We don't bother doing a read/modify/write on the CONFIG register because
>>> + * we only ever use a value of '0' for the other fields in that register.
>>> + */
>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>> +					bool enable)
>>> +{
>>> +	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
>>> +			    IS31FL3216_CONFIG_SSD_DISABLE;
>>> +
>>> +	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
>>> +}
>>> +
>>> +/*
>>> + * NOTE: A mutex is not needed in this function because:
>>> + * - All referenced data is read-only after probe()
>>> + * - The I2C core has a mutex on to protect the bus
>>> + * - There are no read/modify/write operations
>>> + * - Intervening operations between the write of the PWM register
>>> + *   and the Update register are harmless.
>>> + *
>>> + * Example:
>>> + *	PWM_REG_1 write 16
>>> + *	UPDATE_REG write 0
>>> + *	PWM_REG_2 write 128
>>> + *	UPDATE_REG write 0
>>> + *   vs:
>>> + *	PWM_REG_1 write 16
>>> + *	PWM_REG_2 write 128
>>> + *	UPDATE_REG write 0
>>> + *	UPDATE_REG write 0
>>> + * are equivalent. Poking the Update register merely applies all PWM
>>> + * register writes up to that point.
>>> + */
>>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
>>> +				     enum led_brightness brightness)
>>> +{
>>> +	const struct is31fl32xx_led_data *led_data =
>>> +		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
>>> +	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
>>> +	u8 pwm_register_offset;
>>> +	int ret;
>>> +
>>> +	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
>>> +
>>> +	/* NOTE: led_data->channel is 1-based */
>>> +	if (cdef->pwm_registers_reversed)
>>> +		pwm_register_offset = cdef->channels - led_data->channel;
>>> +	else
>>> +		pwm_register_offset = led_data->channel - 1;
>>> +
>>> +	ret = is31fl32xx_write(led_data->priv,
>>> +			       cdef->pwm_register_base + pwm_register_offset,
>>> +			       brightness);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
>>> +}
>>> +
>>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
>>> +{
>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>> +	int ret;
>>> +
>>> +	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
>>> +		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
>>> +		if (ret)
>>> +			return ret;
>>> +	}
>>> +
>>> +	if (cdef->reset_func)
>>> +		return cdef->reset_func(priv);
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
>>> +					bool enable)
>>> +{
>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>> +	int ret;
>>> +
>>> +	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
>>> +		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
>>> +				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
>>> +		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
>>> +		if (ret)
>>> +			return ret;
>>> +	}
>>> +
>>> +	if (cdef->sw_shutdown_func)
>>> +		return cdef->sw_shutdown_func(priv, enable);
>>
>> You seem to call sw_shutdown_func only here, so why should we have
>> enable parameter in this op?
>
> I'm not sure if I understand the question, but I will try to answer.
>
> 'enable' is passed through is31fl32xx_software_shutdown to
> cdef->sw_shutdown_func, so it can be either true or false at that
> point. The purpose of sw_shutdown_func is to add any special behavior
> when enabling/disabling software-shutdown mode, which is needed for
> the 3216 because its SSD bit is in a different position and with
> opposite polarity.
>
> Is it that 'enable' in that line of code makes it look like it's being
> called with an hardcoded value rather than a variable? If so, perhaps a
> different parameter name would make it more obvious? Or a kerneldoc
> comment for the function to describe the parameter?
>
> Or have I totally missed the point of the question?

Actually I should have placed this question next to
the call to s31fl32xx_software_shutdown() in is31fl32xx_init_regs(),
which is passed "false" in the second argument, and there is no
other call to s31fl32xx_software_shutdown() in the driver.

Having the argument makes people wondering that there is some
use case in the driver, where "true" is passed, but it seems not
to be the case.

>>> +	return 0;
>>> +}
>>> +
>>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
>>> +{
>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>> +	int ret;
>>> +
>>> +	ret = is31fl32xx_reset_regs(priv);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	/*
>>> +	 * Set enable bit for all channels.
>>> +	 * We will control state with PWM registers alone.
>>> +	 */
>>> +	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
>>> +		u8 value =
>>> +		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
>>> +		u8 num_regs = cdef->channels /
>>> +				cdef->enable_bits_per_led_control_register;
>>> +		int i;
>>> +
>>> +		for (i = 0; i < num_regs; i++) {
>>> +			ret = is31fl32xx_write(priv,
>>> +					       cdef->led_control_register_base+i,
>>> +					       value);
>>> +			if (ret)
>>> +				return ret;
>>> +		}
>>> +	}
>>> +
>>> +	ret = is31fl32xx_software_shutdown(priv, false);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
>>> +		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
>>> +		if (ret)
>>> +			return ret;
>>> +	}
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
>>> +{
>>> +	return sizeof(struct is31fl32xx_priv) +
>>> +		      (sizeof(struct is31fl32xx_led_data) * num_leds);
>>> +}
>>> +
>>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
>>> +				     const struct device_node *child,
>>> +				     struct is31fl32xx_led_data *led_data)
>>> +{
>>> +	struct led_classdev *cdev = &led_data->cdev;
>>> +	int ret = 0;
>>> +	u32 reg;
>>> +
>>> +	if (of_property_read_string(child, "label", &cdev->name))
>>> +		cdev->name = child->name;
>>> +
>>> +	ret = of_property_read_u32(child, "reg", &reg);
>>> +	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
>>> +		dev_err(dev,
>>> +			"Child node %s does not have a valid reg property\n",
>>> +			child->full_name);
>>> +		return -EINVAL;
>>> +	}
>>> +	led_data->channel = reg;
>>> +
>>> +	of_property_read_string(child, "linux,default-trigger",
>>> +				&cdev->default_trigger);
>>> +
>>> +	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
>>> +					struct is31fl32xx_priv *priv,
>>> +					u8 channel)
>>> +{
>>> +	size_t i;
>>> +
>>> +	for (i = 0; i < priv->num_leds; i++) {
>>> +		if (priv->leds[i].channel == channel)
>>> +			return &priv->leds[i];
>>> +	}
>>> +
>>> +	return NULL;
>>> +}
>>> +
>>> +static int is31fl32xx_parse_dt(struct device *dev,
>>> +			       struct is31fl32xx_priv *priv)
>>> +{
>>> +	struct device_node *child;
>>> +	int ret = 0;
>>> +
>>> +	for_each_child_of_node(dev->of_node, child) {
>>> +		struct is31fl32xx_led_data *led_data =
>>> +			&priv->leds[priv->num_leds];
>>> +		const struct is31fl32xx_led_data *other_led_data;
>>> +
>>> +		led_data->priv = priv;
>>> +
>>> +		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
>>> +		if (ret)
>>> +			goto err;
>>> +
>>> +		/* Detect if channel is already in use by another child */
>>> +		other_led_data = is31fl32xx_find_led_data(priv,
>>> +							  led_data->channel);
>>> +		if (other_led_data) {
>>> +			dev_err(dev,
>>> +				"%s and %s both attempting to use channel %d\n",
>>> +				led_data->cdev.name,
>>> +				other_led_data->cdev.name,
>>> +				led_data->channel);
>>> +			goto err;
>>> +		}
>>> +
>>> +		ret = devm_led_classdev_register(dev, &led_data->cdev);
>>> +		if (ret) {
>>> +			dev_err(dev, "failed to register PWM led for %s: %d\n",
>>> +				led_data->cdev.name, ret);
>>> +			goto err;
>>> +		}
>>> +
>>> +		priv->num_leds++;
>>> +	}
>>> +
>>> +	return 0;
>>> +
>>> +err:
>>> +	of_node_put(child);
>>> +	return ret;
>>> +}
>>> +
>>> +static const struct of_device_id of_is31fl31xx_match[] = {
>>> +	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
>>> +	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
>>> +	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
>>> +	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
>>> +	{},
>>> +};
>>> +
>>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
>>> +
>>> +static int is31fl32xx_probe(struct i2c_client *client,
>>> +			    const struct i2c_device_id *id)
>>> +{
>>> +	const struct is31fl32xx_chipdef *cdef;
>>> +	const struct of_device_id *of_dev_id;
>>> +	struct device *dev = &client->dev;
>>> +	struct is31fl32xx_priv *priv;
>>> +	int count;
>>> +	int ret = 0;
>>> +
>>> +	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
>>> +	if (!of_dev_id)
>>> +		return -EINVAL;
>>> +
>>> +	cdef = of_dev_id->data;
>>> +
>>> +	count = of_get_child_count(dev->of_node);
>>> +	if (!count)
>>> +		return -EINVAL;
>>> +
>>> +	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
>>> +			    GFP_KERNEL);
>>> +	if (!priv)
>>> +		return -ENOMEM;
>>> +
>>> +	priv->client = client;
>>> +	priv->cdef = cdef;
>>> +	i2c_set_clientdata(client, priv);
>>> +
>>> +	ret = is31fl32xx_init_regs(priv);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	ret = is31fl32xx_parse_dt(dev, priv);
>>> +	if (ret)
>>> +		return ret;
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +static int is31fl32xx_remove(struct i2c_client *client)
>>> +{
>>> +	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
>>> +
>>> +	return is31fl32xx_reset_regs(priv);
>>> +}
>>> +
>>> +/*
>>> + * i2c-core requires that id_table be non-NULL, even though
>>> + * it is not used for DeviceTree based instantiation.
>>> + */
>>> +static const struct i2c_device_id is31fl31xx_id[] = {
>>> +	{},
>>> +};
>>> +
>>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
>>> +
>>> +static struct i2c_driver is31fl32xx_driver = {
>>> +	.driver = {
>>> +		.name	= "is31fl32xx",
>>> +		.of_match_table = of_is31fl31xx_match,
>>> +	},
>>> +	.probe		= is31fl32xx_probe,
>>> +	.remove		= is31fl32xx_remove,
>>> +	.id_table	= is31fl31xx_id,
>>> +};
>>> +
>>> +module_i2c_driver(is31fl32xx_driver);
>>> +
>>> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
>>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
>>> +MODULE_LICENSE("GPL v2");
>>>
>
>
>


-- 
Best regards,
Jacek Anaszewski

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[David Rivshin (Allworx)]

(Stefan, sorry for the duplicate, I just realized that I originally 
replied only to you by accident).

On Thu, 3 Mar 2016 19:13:03 +0100 (CET)
Stefan Wahren <stefan.wahren@i2se.com> wrote:

> Hi David,
> 
> i will test the driver on weekend. Some comments below  

Great, thanks very much.

> > "David Rivshin (Allworx)" <drivshin.allworx@gmail.com> hat am 3. März 2016 um
> > 04:01 geschrieben:
> >
> >
> > From: David Rivshin <drivshin@allworx.com>
> >
> > The IS31FL32xx family of LED controllers are I2C devices with multiple
> > constant-current channels, each with independent 256-level PWM control.
> >
> > Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >
> > This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
> > (TI am335x) platform.
> >
> > The programming paradigm of these devices is similar in the following
> > ways:
> > - All registers are 8 bit
> > - All LED control registers are write-only
> > - Each LED channel has a PWM register (0-255)
> > - PWM register writes are shadowed until an Update register is poked
> > - All have a concept of Software Shutdown, which disables output
> >
> > However, there are some differences in devices:
> > - 3236/3235 have a separate Control register for each LED,
> > (3218/3216 pack the enable bits into fewer registers)
> > - 3236/3235 have a per-channel current divisor setting
> > - 3236/3235 have a Global Control register that can turn off all LEDs
> > - 3216 is unique in a number of ways
> > - OUT9-OUT16 can be configured as GPIOs instead of LED controls
> > - LEDs can be programmed with an 8-frame animation, with
> > programmable delay between frames
> > - LEDs can be modulated by an input audio signal
> > - Max output current can be adjusted from 1/4 to 2x globally
> > - Has a Configuration register instead of a Shutdown register
> >
> > This driver currently only supports the base PWM control function
> > of these devices. The following features of these devices are not
> > implemented, although it should be possible to add them in the future:
> > - All devices are capable of going into a lower-power "software
> > shutdown" mode.
> > - The is31fl3236 and is31fl3235 can reduce the max output current
> > per-channel with a divisor of 1, 2, 3, or 4.
> > - The is31fl3216 can use some LED channels as GPIOs instead.
> > - The is31fl3216 can animate LEDs in hardware.
> > - The is31fl3216 can modulate LEDs according to an audio input.
> > - The is31fl3216 can reduce/increase max output current globally.
> >
> > Signed-off-by: David Rivshin <drivshin@allworx.com>
> > ---
> >
> > You may see two instances of this warning:
> > "passing argument 1 of 'of_property_read_string' discards 'const'
> > qualifier from pointer target type"
> > That is a result of of_property_read_string() taking a non-const
> > struct device_node pointer parameter. I have separately submitted a
> > patch to fix that [1], and a few related functions which had the same
> > issue. I'm hoping that will get into linux-next before this does, so
> > that the warnings never show up there.
> >
> > Changes from RFC:
> > - Removed max-brightness DT property.
> > - Refer to these devices as "LED controllers" in Kconfig.
> > - Removed redundant last sentence from Kconfig entry
> > - Removed unnecessary debug code.
> > - Do not set led_classdev.brightness to 0 explicitly, as it is
> > already initialized to 0 by devm_kzalloc().
> > - Used of_property_read_string() instead of of_get_property().
> > - Fail immediately on DT parsing error in a child node, rather than
> > continuing on with the non-faulty ones.
> > - Added additional comments for some things that might be non-obvious.
> > - Added constants for the location of the SSD bit in the SHUTDOWN
> > register, and the 3216's CONFIG register.
> > - Added special sw_shutdown_func for the 3216 device, as that bit
> > is in a different register, at a different position, and has reverse
> > polarity compared to all the other devices.
> > - Refactored is31fl32xx_init_regs() to separate out some logic into
> > is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
> >
> > [1] https://lkml.org/lkml/2016/3/2/746
> >
> > drivers/leds/Kconfig | 8 +
> > drivers/leds/Makefile | 1 +
> > drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
> > 3 files changed, 514 insertions(+)
> > create mode 100644 drivers/leds/leds-is31fl32xx.c
> >
> > diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> > index 1034696..9c63ba4 100644
> > --- a/drivers/leds/Kconfig
> > +++ b/drivers/leds/Kconfig
> > @@ -580,6 +580,14 @@ config LEDS_SN3218
> > This driver can also be built as a module. If so the module
> > will be called leds-sn3218.
> >
> > +config LEDS_IS31FL32XX
> > + tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
> > + depends on LEDS_CLASS && I2C && OF
> > + help
> > + Say Y here to include support for ISSI IS31FL32XX LED controllers.
> > + They are I2C devices with multiple constant-current channels, each
> > + with independent 256-level PWM control.    
> 
> Is it worth to mention the module name here?  

I noticed that some do and some don't. I don't mind adding it, but it
also seemed like it would be obvious, and therefore unnecessary. 

Jacek, which do you prefer?

> > +
> > comment "LED driver for blink(1) USB RGB LED is under Special HID drivers
> > (HID_THINGM)"
> >
> > config LEDS_BLINKM
> > diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
> > index 89c9b6f..3fdf313 100644
> > --- a/drivers/leds/Makefile
> > +++ b/drivers/leds/Makefile
> > @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692) += leds-ktd2692.o
> > obj-$(CONFIG_LEDS_POWERNV) += leds-powernv.o
> > obj-$(CONFIG_LEDS_SEAD3) += leds-sead3.o
> > obj-$(CONFIG_LEDS_SN3218) += leds-sn3218.o
> > +obj-$(CONFIG_LEDS_IS31FL32XX) += leds-is31fl32xx.o
> >
> > # LED SPI Drivers
> > obj-$(CONFIG_LEDS_DAC124S085) += leds-dac124s085.o
> > diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
> > new file mode 100644
> > index 0000000..49818f0
> > --- /dev/null
> > +++ b/drivers/leds/leds-is31fl32xx.c
> > @@ -0,0 +1,505 @@
> > +/*
> > + * linux/drivers/leds-is31fl32xx.c    
> 
> I think this is unnecessary.  

I tend to agree. I think I used leds-pwm.c as a template, and that had 
such a comment. I assumed it was coding-style and kept it, but now I see 
that only a minority of led drivers have it. If I do another spin for 
any reason I'll remove it.
 
> > + *
> > + * Driver for ISSI IS31FL32xx family of I2C LED controllers
> > + *
> > + * Copyright 2015 Allworx Corp.
> > + *
> > + *
> > + * 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.
> > + *
> > + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> > + */
> > +    
> 
> Shouldn't we include <linux/device.h> here?  

Good catch. I was getting that via i2c.h, but since struct device is
referenced explicitly in a few places, device.h should probably be
included directly.

> > +#include <linux/err.h>
> > +#include <linux/i2c.h>
> > +#include <linux/kernel.h>
> > +#include <linux/leds.h>
> > +#include <linux/module.h>
> > +#include <linux/of_platform.h>
> > +
> > +/* Used to indicate a device has no such register */
> > +#define IS31FL32XX_REG_NONE 0xFF
> > +
> > +/* Software Shutdown bit in Shutdown Register */
> > +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE 0
> > +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
> > +
> > +/* IS31FL3216 has a number of unique registers */
> > +#define IS31FL3216_CONFIG_REG 0x00
> > +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
> > +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
> > +
> > +/* Software Shutdown bit in 3216 Config Register */
> > +#define IS31FL3216_CONFIG_SSD_ENABLE BIT(7)
> > +#define IS31FL3216_CONFIG_SSD_DISABLE 0
> > +
> > +struct is31fl32xx_priv;
> > +struct is31fl32xx_led_data {
> > + struct led_classdev cdev;
> > + u8 channel; /* 1-based, max priv->cdef->channels */
> > + struct is31fl32xx_priv *priv;
> > +};
> > +
> > +struct is31fl32xx_priv {
> > + const struct is31fl32xx_chipdef *cdef;
> > + struct i2c_client *client;
> > + unsigned int num_leds;
> > + struct is31fl32xx_led_data leds[0];
> > +};
> > +
> > +/**
> > + * struct is31fl32xx_chipdef - chip-specific attributes
> > + * @channels : Number of LED channels
> > + * @shutdown_reg : address of Shutdown register (optional)
> > + * @pwm_update_reg : address of PWM Update register
> > + * @global_control_reg : address of Global Control register (optional)
> > + * @reset_reg : address of Reset register (optional)
> > + * @pwm_register_base : address of first PWM register
> > + * @pwm_registers_reversed: : true if PWM registers count down instead of up
> > + * @led_control_register_base : address of first LED control register
> > (optional)
> > + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
> > + * @reset_func: : pointer to reset function
> > + *
> > + * For all optional register addresses, the sentinel value
> > %IS31FL32XX_REG_NONE
> > + * indicates that this chip has no such register.
> > + *
> > + * If non-NULL, @reset_func will be called during probing to set all
> > + * necessary registers to a known initialization state. This is needed
> > + * for chips that do not have a @reset_reg.
> > + *
> > + * @enable_bits_per_led_control_register must be >=1 if
> > + * @led_control_register_base != %IS31FL32XX_REG_NONE.
> > + */
> > +struct is31fl32xx_chipdef {
> > + u8 channels;
> > + u8 shutdown_reg;
> > + u8 pwm_update_reg;
> > + u8 global_control_reg;
> > + u8 reset_reg;
> > + u8 pwm_register_base;
> > + bool pwm_registers_reversed;
> > + u8 led_control_register_base;
> > + u8 enable_bits_per_led_control_register;
> > + int (*reset_func)(struct is31fl32xx_priv *priv);
> > + int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
> > +};
> > +
> > +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
> > + .channels = 36,
> > + .shutdown_reg = 0x00,
> > + .pwm_update_reg = 0x25,
> > + .global_control_reg = 0x4a,
> > + .reset_reg = 0x4f,
> > + .pwm_register_base = 0x01,
> > + .led_control_register_base = 0x26,
> > + .enable_bits_per_led_control_register = 1,
> > +};
> > +
> > +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
> > + .channels = 28,
> > + .shutdown_reg = 0x00,
> > + .pwm_update_reg = 0x25,
> > + .global_control_reg = 0x4a,
> > + .reset_reg = 0x4f,
> > + .pwm_register_base = 0x05,
> > + .led_control_register_base = 0x2a,
> > + .enable_bits_per_led_control_register = 1,
> > +};
> > +
> > +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
> > + .channels = 18,
> > + .shutdown_reg = 0x00,
> > + .pwm_update_reg = 0x16,
> > + .global_control_reg = IS31FL32XX_REG_NONE,
> > + .reset_reg = 0x17,
> > + .pwm_register_base = 0x01,
> > + .led_control_register_base = 0x13,
> > + .enable_bits_per_led_control_register = 6,
> > +};
> > +
> > +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
> > +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> > + bool enable);
> > +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
> > + .channels = 16,
> > + .shutdown_reg = IS31FL32XX_REG_NONE,
> > + .pwm_update_reg = 0xB0,
> > + .global_control_reg = IS31FL32XX_REG_NONE,
> > + .reset_reg = IS31FL32XX_REG_NONE,
> > + .pwm_register_base = 0x10,
> > + .pwm_registers_reversed = true,
> > + .led_control_register_base = 0x01,
> > + .enable_bits_per_led_control_register = 8,
> > + .reset_func = is31fl3216_reset,
> > + .sw_shutdown_func = is31fl3216_software_shutdown,
> > +};
> > +
> > +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
> > +{
> > + int ret;
> > +
> > + dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
> > +
> > + ret = i2c_smbus_write_byte_data(priv->client, reg, val);
> > + if (ret) {
> > + dev_err(&priv->client->dev,
> > + "register write to 0x%02X failed (error %d)",
> > + reg, ret);
> > + }    
> 
> In case somebody use this driver as heartbeat and writing fails permanently the
> log will be flooded.  

Unless I'm mistaken that would require the device/bus to fail after 
successfully probing (probe code itself bails on the first write 
failure, so there would be no flooding as a result of that). So while 
not impossible, I imagine it would be unlikely, and I'd hate to remove 
an error message for such an important condition. 

I suppose I could use dev_err_ratelimited() to soften any potential 
flooding, but I second guess that because:
 - In led_core.c set_brightness_delayed() has a dev_err() that would come 
   out on each failed LED update anyways.
 - There is precedent in other led drivers of a similar error message.
 - Some userspace logging programs will compresses repeated messages anyways.

Jacek, what is your preference on this? 

> > + return ret;
> > +}
> > +
> > +/*
> > + * Custom reset function for IS31FL3216 because it does not have a RESET
> > + * register the way that the other IS31FL32xx chips do. We don't bother
> > + * writing the GPIO and animation registers, because the registers we
> > + * do write ensure those will have no effect.
> > + */
> > +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
> > +{
> > + unsigned int i;
> > + int ret;
> > +
> > + ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
> > + IS31FL3216_CONFIG_SSD_ENABLE);
> > + if (ret)
> > + return ret;
> > + for (i = 0; i < priv->cdef->channels; i++) {
> > + ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
> > + 0x00);
> > + if (ret)
> > + return ret;
> > + }
> > + ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
> > + if (ret)
> > + return ret;
> > + ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
> > + if (ret)
> > + return ret;
> > + ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
> > + if (ret)
> > + return ret;
> > +
> > + return 0;
> > +}
> > +
> > +/*
> > + * Custom Software-Shutdown function for IS31FL3216 because it does not have
> > + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
> > + * We don't bother doing a read/modify/write on the CONFIG register because
> > + * we only ever use a value of '0' for the other fields in that register.
> > + */
> > +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> > + bool enable)
> > +{
> > + u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
> > + IS31FL3216_CONFIG_SSD_DISABLE;
> > +
> > + return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
> > +}
> > +
> > +/*
> > + * NOTE: A mutex is not needed in this function because:
> > + * - All referenced data is read-only after probe()
> > + * - The I2C core has a mutex on to protect the bus
> > + * - There are no read/modify/write operations
> > + * - Intervening operations between the write of the PWM register
> > + * and the Update register are harmless.
> > + *
> > + * Example:
> > + * PWM_REG_1 write 16
> > + * UPDATE_REG write 0
> > + * PWM_REG_2 write 128
> > + * UPDATE_REG write 0
> > + * vs:
> > + * PWM_REG_1 write 16
> > + * PWM_REG_2 write 128
> > + * UPDATE_REG write 0
> > + * UPDATE_REG write 0
> > + * are equivalent. Poking the Update register merely applies all PWM
> > + * register writes up to that point.
> > + */
> > +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
> > + enum led_brightness brightness)
> > +{
> > + const struct is31fl32xx_led_data *led_data =
> > + container_of(led_cdev, struct is31fl32xx_led_data, cdev);
> > + const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
> > + u8 pwm_register_offset;
> > + int ret;
> > +
> > + dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
> > +
> > + /* NOTE: led_data->channel is 1-based */
> > + if (cdef->pwm_registers_reversed)
> > + pwm_register_offset = cdef->channels - led_data->channel;
> > + else
> > + pwm_register_offset = led_data->channel - 1;
> > +
> > + ret = is31fl32xx_write(led_data->priv,
> > + cdef->pwm_register_base + pwm_register_offset,
> > + brightness);
> > + if (ret)
> > + return ret;
> > +
> > + return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
> > +}
> > +
> > +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
> > +{
> > + const struct is31fl32xx_chipdef *cdef = priv->cdef;
> > + int ret;
> > +
> > + if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
> > + ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
> > + if (ret)
> > + return ret;
> > + }
> > +
> > + if (cdef->reset_func)
> > + return cdef->reset_func(priv);
> > +
> > + return 0;
> > +}
> > +
> > +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
> > + bool enable)
> > +{
> > + const struct is31fl32xx_chipdef *cdef = priv->cdef;
> > + int ret;
> > +
> > + if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
> > + u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
> > + IS31FL32XX_SHUTDOWN_SSD_DISABLE;
> > + ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
> > + if (ret)
> > + return ret;
> > + }
> > +
> > + if (cdef->sw_shutdown_func)
> > + return cdef->sw_shutdown_func(priv, enable);
> > +
> > + return 0;
> > +}
> > +
> > +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
> > +{
> > + const struct is31fl32xx_chipdef *cdef = priv->cdef;
> > + int ret;
> > +
> > + ret = is31fl32xx_reset_regs(priv);
> > + if (ret)
> > + return ret;
> > +
> > + /*
> > + * Set enable bit for all channels.
> > + * We will control state with PWM registers alone.
> > + */
> > + if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
> > + u8 value =
> > + GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
> > + u8 num_regs = cdef->channels /
> > + cdef->enable_bits_per_led_control_register;
> > + int i;
> > +
> > + for (i = 0; i < num_regs; i++) {
> > + ret = is31fl32xx_write(priv,
> > + cdef->led_control_register_base+i,
> > + value);
> > + if (ret)
> > + return ret;
> > + }
> > + }
> > +
> > + ret = is31fl32xx_software_shutdown(priv, false);
> > + if (ret)
> > + return ret;
> > +
> > + if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
> > + ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
> > + if (ret)
> > + return ret;
> > + }
> > +
> > + return 0;
> > +}
> > +
> > +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
> > +{
> > + return sizeof(struct is31fl32xx_priv) +
> > + (sizeof(struct is31fl32xx_led_data) * num_leds);
> > +}
> > +
> > +static int is31fl32xx_parse_child_dt(const struct device *dev,
> > + const struct device_node *child,
> > + struct is31fl32xx_led_data *led_data)
> > +{
> > + struct led_classdev *cdev = &led_data->cdev;
> > + int ret = 0;
> > + u32 reg;
> > +
> > + if (of_property_read_string(child, "label", &cdev->name))
> > + cdev->name = child->name;
> > +
> > + ret = of_property_read_u32(child, "reg", &reg);
> > + if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
> > + dev_err(dev,
> > + "Child node %s does not have a valid reg property\n",
> > + child->full_name);
> > + return -EINVAL;
> > + }
> > + led_data->channel = reg;
> > +
> > + of_property_read_string(child, "linux,default-trigger",
> > + &cdev->default_trigger);
> > +
> > + cdev->brightness_set_blocking = is31fl32xx_brightness_set;
> > +
> > + return 0;
> > +}
> > +
> > +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
> > + struct is31fl32xx_priv *priv,
> > + u8 channel)
> > +{
> > + size_t i;
> > +
> > + for (i = 0; i < priv->num_leds; i++) {
> > + if (priv->leds[i].channel == channel)
> > + return &priv->leds[i];
> > + }
> > +
> > + return NULL;
> > +}
> > +
> > +static int is31fl32xx_parse_dt(struct device *dev,
> > + struct is31fl32xx_priv *priv)
> > +{
> > + struct device_node *child;
> > + int ret = 0;
> > +
> > + for_each_child_of_node(dev->of_node, child) {
> > + struct is31fl32xx_led_data *led_data =
> > + &priv->leds[priv->num_leds];    
> 
> Maybe i missed something, but is it really protected against out of index
> access?  

The array is allocated with size equal to the number of child nodes,
and num_leds is incremented once for each child node parsed. So in 
order for the index to be out of bounds, the number of child nodes
would need to increase during the probe. I assumed that the DT is 
static during probing, but if that's not the case then you're right 
that this is a potential problem. Also, this equivalent logic is 
used in leds-pwm, leds-gpio, and leds-ns2, so that gives me 
confidence that its safe. 
Unless DT overlays change that assumption? 

> > + const struct is31fl32xx_led_data *other_led_data;
> > +
> > + led_data->priv = priv;
> > +
> > + ret = is31fl32xx_parse_child_dt(dev, child, led_data);
> > + if (ret)
> > + goto err;
> > +
> > + /* Detect if channel is already in use by another child */
> > + other_led_data = is31fl32xx_find_led_data(priv,
> > + led_data->channel);
> > + if (other_led_data) {
> > + dev_err(dev,
> > + "%s and %s both attempting to use channel %d\n",
> > + led_data->cdev.name,
> > + other_led_data->cdev.name,
> > + led_data->channel);
> > + goto err;
> > + }
> > +
> > + ret = devm_led_classdev_register(dev, &led_data->cdev);
> > + if (ret) {
> > + dev_err(dev, "failed to register PWM led for %s: %d\n",
> > + led_data->cdev.name, ret);
> > + goto err;
> > + }
> > +
> > + priv->num_leds++;
> > + }
> > +
> > + return 0;
> > +
> > +err:
> > + of_node_put(child);
> > + return ret;
> > +}
> > +
> > +static const struct of_device_id of_is31fl31xx_match[] = {
> > + { .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
> > + { .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
> > + { .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
> > + { .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
> > + {},
> > +};
> > +
> > +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
> > +
> > +static int is31fl32xx_probe(struct i2c_client *client,
> > + const struct i2c_device_id *id)
> > +{
> > + const struct is31fl32xx_chipdef *cdef;
> > + const struct of_device_id *of_dev_id;
> > + struct device *dev = &client->dev;
> > + struct is31fl32xx_priv *priv;
> > + int count;
> > + int ret = 0;
> > +
> > + of_dev_id = of_match_device(of_is31fl31xx_match, dev);
> > + if (!of_dev_id)
> > + return -EINVAL;
> > +
> > + cdef = of_dev_id->data;
> > +
> > + count = of_get_child_count(dev->of_node);
> > + if (!count)
> > + return -EINVAL;
> > +
> > + priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
> > + GFP_KERNEL);
> > + if (!priv)
> > + return -ENOMEM;
> > +
> > + priv->client = client;
> > + priv->cdef = cdef;
> > + i2c_set_clientdata(client, priv);
> > +
> > + ret = is31fl32xx_init_regs(priv);
> > + if (ret)
> > + return ret;
> > +
> > + ret = is31fl32xx_parse_dt(dev, priv);
> > + if (ret)
> > + return ret;
> > +
> > + return 0;
> > +}
> > +
> > +static int is31fl32xx_remove(struct i2c_client *client)
> > +{
> > + struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
> > +
> > + return is31fl32xx_reset_regs(priv);
> > +}
> > +
> > +/*
> > + * i2c-core requires that id_table be non-NULL, even though
> > + * it is not used for DeviceTree based instantiation.
> > + */
> > +static const struct i2c_device_id is31fl31xx_id[] = {
> > + {},
> > +};
> > +
> > +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
> > +
> > +static struct i2c_driver is31fl32xx_driver = {
> > + .driver = {
> > + .name = "is31fl32xx",
> > + .of_match_table = of_is31fl31xx_match,
> > + },
> > + .probe = is31fl32xx_probe,
> > + .remove = is31fl32xx_remove,    
> 
> Sorry, what was the reason to skip shutdown?  

If I understood Jacek's last email on the topic [1] correctly, he's now 
of the opinion that the decision to turn LEDs off on reboot should be 
left to userspace, rather than done by the driver. For these devices, 
the only thing a shutdown callback would do is turn off the LEDs (through 
any of multiple methods). So, if we want to leave the state as-is on
reboot there's no need for a shutdown callback.

[1] http://www.spinics.net/lists/linux-leds/msg05644.html

> > + .id_table = is31fl31xx_id,
> > +};
> > +
> > +module_i2c_driver(is31fl32xx_driver);
> > +
> > +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
> > +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
> > +MODULE_LICENSE("GPL v2");
> > --
> > 2.5.0
> >    

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[David Rivshin (Allworx)]

On Fri, 04 Mar 2016 08:54:02 +0100
Jacek Anaszewski <j.anaszewski-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org> wrote:

> On 03/04/2016 01:45 AM, David Rivshin (Allworx) wrote:
> > On Thu, 03 Mar 2016 15:51:32 +0100
> > Jacek Anaszewski <j.anaszewski-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org> wrote:
> >  
> >> Hi David,
> >>
> >> Thanks for the update. Two remarks in the code.
> >>
> >> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:  
> >>> From: David Rivshin <drivshin-5fOYsn7Fw8lBDgjK7y7TUQ@public.gmane.org>
> >>>
> >>> The IS31FL32xx family of LED controllers are I2C devices with multiple
> >>> constant-current channels, each with independent 256-level PWM control.
> >>>
> >>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >>>
> >>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
> >>> (TI am335x) platform.
> >>>
> >>> The programming paradigm of these devices is similar in the following
> >>> ways:
> >>>    - All registers are 8 bit
> >>>    - All LED control registers are write-only
> >>>    - Each LED channel has a PWM register (0-255)
> >>>    - PWM register writes are shadowed until an Update register is poked
> >>>    - All have a concept of Software Shutdown, which disables output
> >>>
> >>> However, there are some differences in devices:
> >>>    - 3236/3235 have a separate Control register for each LED,
> >>>      (3218/3216 pack the enable bits into fewer registers)
> >>>    - 3236/3235 have a per-channel current divisor setting
> >>>    - 3236/3235 have a Global Control register that can turn off all LEDs
> >>>    - 3216 is unique in a number of ways
> >>>       - OUT9-OUT16 can be configured as GPIOs instead of LED controls
> >>>       - LEDs can be programmed with an 8-frame animation, with
> >>>         programmable delay between frames
> >>>       - LEDs can be modulated by an input audio signal
> >>>       - Max output current can be adjusted from 1/4 to 2x globally
> >>>       - Has a Configuration register instead of a Shutdown register
> >>>
> >>> This driver currently only supports the base PWM control function
> >>> of these devices. The following features of these devices are not
> >>> implemented, although it should be possible to add them in the future:
> >>>    - All devices are capable of going into a lower-power "software
> >>>      shutdown" mode.
> >>>    - The is31fl3236 and is31fl3235 can reduce the max output current
> >>>      per-channel with a divisor of 1, 2, 3, or 4.
> >>>    - The is31fl3216 can use some LED channels as GPIOs instead.
> >>>    - The is31fl3216 can animate LEDs in hardware.
> >>>    - The is31fl3216 can modulate LEDs according to an audio input.
> >>>    - The is31fl3216 can reduce/increase max output current globally.
> >>>
> >>> Signed-off-by: David Rivshin <drivshin-5fOYsn7Fw8lBDgjK7y7TUQ@public.gmane.org>
> >>> ---
> >>>
> >>> You may see two instances of this warning:
> >>>     "passing argument 1 of 'of_property_read_string' discards 'const'
> >>>      qualifier from pointer target type"
> >>> That is a result of of_property_read_string() taking a non-const
> >>> struct device_node pointer parameter. I have separately submitted a
> >>> patch to fix that [1], and a few related functions which had the same
> >>> issue. I'm hoping that will get into linux-next before this does, so
> >>> that the warnings never show up there.  
> >>
> >> Please adjust the patch so that it compiles without warnings on
> >> current linux-next. Your patch for DT API hasn't been reviewed yet
> >> AFICS, and I can imagine that there will be some resistance against.  
> >
> > Since the DT API patch was just accepted by Rob [1], would it be OK
> > to wait for the results of Stefan's testing (and any other reviews)
> > before making a decision on this? From Stefan's note, it won't be
> > until this weekend that he will have a chance to test, and I'm
> > guessing the DT API patch will make its way through Rob's tree to
> > linux-next by then.  
> 
> OK.
> 
> > FYI, the warning workaround would be to make the second parameter to
> > is31fl32xx_parse_child_dt() non-const.
> >
> > [1] https://lkml.org/lkml/2016/3/3/924
> >  
> >>> Changes from RFC:
> >>>    - Removed max-brightness DT property.
> >>>    - Refer to these devices as "LED controllers" in Kconfig.
> >>>    - Removed redundant last sentence from Kconfig entry
> >>>    - Removed unnecessary debug code.
> >>>    - Do not set led_classdev.brightness to 0 explicitly, as it is
> >>>      already initialized to 0 by devm_kzalloc().
> >>>    - Used of_property_read_string() instead of of_get_property().
> >>>    - Fail immediately on DT parsing error in a child node, rather than
> >>>      continuing on with the non-faulty ones.
> >>>    - Added additional comments for some things that might be non-obvious.
> >>>    - Added constants for the location of the SSD bit in the SHUTDOWN
> >>>      register, and the 3216's CONFIG register.
> >>>    - Added special sw_shutdown_func for the 3216 device, as that bit
> >>>      is in a different register, at a different position, and has reverse
> >>>      polarity compared to all the other devices.
> >>>    - Refactored is31fl32xx_init_regs() to separate out some logic into
> >>>      is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
> >>>
> >>> [1] https://lkml.org/lkml/2016/3/2/746
> >>>
> >>>    drivers/leds/Kconfig           |   8 +
> >>>    drivers/leds/Makefile          |   1 +
> >>>    drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
> >>>    3 files changed, 514 insertions(+)
> >>>    create mode 100644 drivers/leds/leds-is31fl32xx.c
> >>>
> >>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> >>> index 1034696..9c63ba4 100644
> >>> --- a/drivers/leds/Kconfig
> >>> +++ b/drivers/leds/Kconfig
> >>> @@ -580,6 +580,14 @@ config LEDS_SN3218
> >>>    	  This driver can also be built as a module. If so the module
> >>>    	  will be called leds-sn3218.
> >>>
> >>> +config LEDS_IS31FL32XX
> >>> +	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
> >>> +	depends on LEDS_CLASS && I2C && OF
> >>> +	help
> >>> +	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
> >>> +	  They are I2C devices with multiple constant-current channels, each
> >>> +	  with independent 256-level PWM control.
> >>> +
> >>>    comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
> >>>
> >>>    config LEDS_BLINKM
> >>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
> >>> index 89c9b6f..3fdf313 100644
> >>> --- a/drivers/leds/Makefile
> >>> +++ b/drivers/leds/Makefile
> >>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
> >>>    obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
> >>>    obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
> >>>    obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
> >>> +obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
> >>>
> >>>    # LED SPI Drivers
> >>>    obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
> >>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
> >>> new file mode 100644
> >>> index 0000000..49818f0
> >>> --- /dev/null
> >>> +++ b/drivers/leds/leds-is31fl32xx.c
> >>> @@ -0,0 +1,505 @@
> >>> +/*
> >>> + * linux/drivers/leds-is31fl32xx.c
> >>> + *
> >>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
> >>> + *
> >>> + * Copyright 2015 Allworx Corp.
> >>> + *
> >>> + *
> >>> + * 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.
> >>> + *
> >>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >>> + */
> >>> +
> >>> +#include <linux/err.h>
> >>> +#include <linux/i2c.h>
> >>> +#include <linux/kernel.h>
> >>> +#include <linux/leds.h>
> >>> +#include <linux/module.h>
> >>> +#include <linux/of_platform.h>
> >>> +
> >>> +/* Used to indicate a device has no such register */
> >>> +#define IS31FL32XX_REG_NONE 0xFF
> >>> +
> >>> +/* Software Shutdown bit in Shutdown Register */
> >>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
> >>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
> >>> +
> >>> +/* IS31FL3216 has a number of unique registers */
> >>> +#define IS31FL3216_CONFIG_REG 0x00
> >>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
> >>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
> >>> +
> >>> +/* Software Shutdown bit in 3216 Config Register */
> >>> +#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
> >>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
> >>> +
> >>> +struct is31fl32xx_priv;
> >>> +struct is31fl32xx_led_data {
> >>> +	struct led_classdev cdev;
> >>> +	u8 channel; /* 1-based, max priv->cdef->channels */
> >>> +	struct is31fl32xx_priv *priv;
> >>> +};
> >>> +
> >>> +struct is31fl32xx_priv {
> >>> +	const struct is31fl32xx_chipdef *cdef;
> >>> +	struct i2c_client *client;
> >>> +	unsigned int num_leds;
> >>> +	struct is31fl32xx_led_data leds[0];
> >>> +};
> >>> +
> >>> +/**
> >>> + * struct is31fl32xx_chipdef - chip-specific attributes
> >>> + * @channels            : Number of LED channels
> >>> + * @shutdown_reg        : address of Shutdown register (optional)
> >>> + * @pwm_update_reg      : address of PWM Update register
> >>> + * @global_control_reg  : address of Global Control register (optional)
> >>> + * @reset_reg           : address of Reset register (optional)
> >>> + * @pwm_register_base   : address of first PWM register
> >>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
> >>> + * @led_control_register_base : address of first LED control register (optional)
> >>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
> >>> + * @reset_func:         : pointer to reset function
> >>> + *
> >>> + * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
> >>> + * indicates that this chip has no such register.
> >>> + *
> >>> + * If non-NULL, @reset_func will be called during probing to set all
> >>> + * necessary registers to a known initialization state. This is needed
> >>> + * for chips that do not have a @reset_reg.
> >>> + *
> >>> + * @enable_bits_per_led_control_register must be >=1 if
> >>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
> >>> + */
> >>> +struct is31fl32xx_chipdef {
> >>> +	u8	channels;
> >>> +	u8	shutdown_reg;
> >>> +	u8	pwm_update_reg;
> >>> +	u8	global_control_reg;
> >>> +	u8	reset_reg;
> >>> +	u8	pwm_register_base;
> >>> +	bool	pwm_registers_reversed;
> >>> +	u8	led_control_register_base;
> >>> +	u8	enable_bits_per_led_control_register;
> >>> +	int (*reset_func)(struct is31fl32xx_priv *priv);
> >>> +	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
> >>> +};
> >>> +
> >>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
> >>> +	.channels				= 36,
> >>> +	.shutdown_reg				= 0x00,
> >>> +	.pwm_update_reg				= 0x25,
> >>> +	.global_control_reg			= 0x4a,
> >>> +	.reset_reg				= 0x4f,
> >>> +	.pwm_register_base			= 0x01,
> >>> +	.led_control_register_base		= 0x26,
> >>> +	.enable_bits_per_led_control_register	= 1,
> >>> +};
> >>> +
> >>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
> >>> +	.channels				= 28,
> >>> +	.shutdown_reg				= 0x00,
> >>> +	.pwm_update_reg				= 0x25,
> >>> +	.global_control_reg			= 0x4a,
> >>> +	.reset_reg				= 0x4f,
> >>> +	.pwm_register_base			= 0x05,
> >>> +	.led_control_register_base		= 0x2a,
> >>> +	.enable_bits_per_led_control_register	= 1,
> >>> +};
> >>> +
> >>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
> >>> +	.channels				= 18,
> >>> +	.shutdown_reg				= 0x00,
> >>> +	.pwm_update_reg				= 0x16,
> >>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
> >>> +	.reset_reg				= 0x17,
> >>> +	.pwm_register_base			= 0x01,
> >>> +	.led_control_register_base		= 0x13,
> >>> +	.enable_bits_per_led_control_register	= 6,
> >>> +};
> >>> +
> >>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
> >>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> >>> +					bool enable);
> >>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
> >>> +	.channels				= 16,
> >>> +	.shutdown_reg				= IS31FL32XX_REG_NONE,
> >>> +	.pwm_update_reg				= 0xB0,
> >>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
> >>> +	.reset_reg				= IS31FL32XX_REG_NONE,
> >>> +	.pwm_register_base			= 0x10,
> >>> +	.pwm_registers_reversed			= true,
> >>> +	.led_control_register_base		= 0x01,
> >>> +	.enable_bits_per_led_control_register	= 8,
> >>> +	.reset_func				= is31fl3216_reset,
> >>> +	.sw_shutdown_func			= is31fl3216_software_shutdown,
> >>> +};
> >>> +
> >>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
> >>> +{
> >>> +	int ret;
> >>> +
> >>> +	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
> >>> +
> >>> +	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
> >>> +	if (ret) {
> >>> +		dev_err(&priv->client->dev,
> >>> +			"register write to 0x%02X failed (error %d)",
> >>> +			reg, ret);
> >>> +	}
> >>> +	return ret;
> >>> +}
> >>> +
> >>> +/*
> >>> + * Custom reset function for IS31FL3216 because it does not have a RESET
> >>> + * register the way that the other IS31FL32xx chips do. We don't bother
> >>> + * writing the GPIO and animation registers, because the registers we
> >>> + * do write ensure those will have no effect.
> >>> + */
> >>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
> >>> +{
> >>> +	unsigned int i;
> >>> +	int ret;
> >>> +
> >>> +	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
> >>> +			       IS31FL3216_CONFIG_SSD_ENABLE);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +	for (i = 0; i < priv->cdef->channels; i++) {
> >>> +		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
> >>> +				       0x00);
> >>> +		if (ret)
> >>> +			return ret;
> >>> +	}
> >>> +	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	return 0;
> >>> +}
> >>> +
> >>> +/*
> >>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
> >>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
> >>> + * We don't bother doing a read/modify/write on the CONFIG register because
> >>> + * we only ever use a value of '0' for the other fields in that register.
> >>> + */
> >>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> >>> +					bool enable)
> >>> +{
> >>> +	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
> >>> +			    IS31FL3216_CONFIG_SSD_DISABLE;
> >>> +
> >>> +	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
> >>> +}
> >>> +
> >>> +/*
> >>> + * NOTE: A mutex is not needed in this function because:
> >>> + * - All referenced data is read-only after probe()
> >>> + * - The I2C core has a mutex on to protect the bus
> >>> + * - There are no read/modify/write operations
> >>> + * - Intervening operations between the write of the PWM register
> >>> + *   and the Update register are harmless.
> >>> + *
> >>> + * Example:
> >>> + *	PWM_REG_1 write 16
> >>> + *	UPDATE_REG write 0
> >>> + *	PWM_REG_2 write 128
> >>> + *	UPDATE_REG write 0
> >>> + *   vs:
> >>> + *	PWM_REG_1 write 16
> >>> + *	PWM_REG_2 write 128
> >>> + *	UPDATE_REG write 0
> >>> + *	UPDATE_REG write 0
> >>> + * are equivalent. Poking the Update register merely applies all PWM
> >>> + * register writes up to that point.
> >>> + */
> >>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
> >>> +				     enum led_brightness brightness)
> >>> +{
> >>> +	const struct is31fl32xx_led_data *led_data =
> >>> +		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
> >>> +	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
> >>> +	u8 pwm_register_offset;
> >>> +	int ret;
> >>> +
> >>> +	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
> >>> +
> >>> +	/* NOTE: led_data->channel is 1-based */
> >>> +	if (cdef->pwm_registers_reversed)
> >>> +		pwm_register_offset = cdef->channels - led_data->channel;
> >>> +	else
> >>> +		pwm_register_offset = led_data->channel - 1;
> >>> +
> >>> +	ret = is31fl32xx_write(led_data->priv,
> >>> +			       cdef->pwm_register_base + pwm_register_offset,
> >>> +			       brightness);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
> >>> +}
> >>> +
> >>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
> >>> +{
> >>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>> +	int ret;
> >>> +
> >>> +	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
> >>> +		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
> >>> +		if (ret)
> >>> +			return ret;
> >>> +	}
> >>> +
> >>> +	if (cdef->reset_func)
> >>> +		return cdef->reset_func(priv);
> >>> +
> >>> +	return 0;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
> >>> +					bool enable)
> >>> +{
> >>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>> +	int ret;
> >>> +
> >>> +	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
> >>> +		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
> >>> +				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
> >>> +		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
> >>> +		if (ret)
> >>> +			return ret;
> >>> +	}
> >>> +
> >>> +	if (cdef->sw_shutdown_func)
> >>> +		return cdef->sw_shutdown_func(priv, enable);  
> >>
> >> You seem to call sw_shutdown_func only here, so why should we have
> >> enable parameter in this op?  
> >
> > I'm not sure if I understand the question, but I will try to answer.
> >
> > 'enable' is passed through is31fl32xx_software_shutdown to
> > cdef->sw_shutdown_func, so it can be either true or false at that
> > point. The purpose of sw_shutdown_func is to add any special behavior
> > when enabling/disabling software-shutdown mode, which is needed for
> > the 3216 because its SSD bit is in a different position and with
> > opposite polarity.
> >
> > Is it that 'enable' in that line of code makes it look like it's being
> > called with an hardcoded value rather than a variable? If so, perhaps a
> > different parameter name would make it more obvious? Or a kerneldoc
> > comment for the function to describe the parameter?
> >
> > Or have I totally missed the point of the question?  
> 
> Actually I should have placed this question next to
> the call to s31fl32xx_software_shutdown() in is31fl32xx_init_regs(),
> which is passed "false" in the second argument, and there is no
> other call to s31fl32xx_software_shutdown() in the driver.
> 
> Having the argument makes people wondering that there is some
> use case in the driver, where "true" is passed, but it seems not
> to be the case.

Thanks for the clarification.

Yes, there is currently no explicit call to enable software-shutdown
mode. Since the reset state of these devices is to have software-shutdown 
enabled, the only explicit operation on it that's currently needed is
to disable it.

Reasons I can think to have the 'enable' parameter anyways include:
- It seems logical to me that an API to manipulate the software-shutdown
  state should allow both enabling and disabling. 
- Having a parameter for that seemed the most obvious way to go, and it
  was trivial to implement. 
- Alternatively having separate "enable" and "disable" functions would 
  duplicate most of the logic, vs the parameter being just a single 
  conditional. And that would also imply two function pointers in the 
  chipdefs, which I'd prefer to minimize.
- If anyone wanted to implement suspend/resume in the future, they would
  most likely do it by enabling/disabling software-shutdown. Supporting
  both enable/disable from the start should make that trivial to do.
- I thought the code read better with a bool parameter, vs a longer 
  function name.

So nothing really critical, but mostly just my aesthetic and preference.

Also, I expect that is31fl32xx_software_shutdown() would be inlined, so 
the conditional check in there is optimized out anyways, and there is no 
performance penalty. Looking at the disassembly in my ARMv7a build, both 
of those have indeed happened there.

> >>> +	return 0;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
> >>> +{
> >>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>> +	int ret;
> >>> +
> >>> +	ret = is31fl32xx_reset_regs(priv);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	/*
> >>> +	 * Set enable bit for all channels.
> >>> +	 * We will control state with PWM registers alone.
> >>> +	 */
> >>> +	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
> >>> +		u8 value =
> >>> +		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
> >>> +		u8 num_regs = cdef->channels /
> >>> +				cdef->enable_bits_per_led_control_register;
> >>> +		int i;
> >>> +
> >>> +		for (i = 0; i < num_regs; i++) {
> >>> +			ret = is31fl32xx_write(priv,
> >>> +					       cdef->led_control_register_base+i,
> >>> +					       value);
> >>> +			if (ret)
> >>> +				return ret;
> >>> +		}
> >>> +	}
> >>> +
> >>> +	ret = is31fl32xx_software_shutdown(priv, false);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
> >>> +		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
> >>> +		if (ret)
> >>> +			return ret;
> >>> +	}
> >>> +
> >>> +	return 0;
> >>> +}
> >>> +
> >>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
> >>> +{
> >>> +	return sizeof(struct is31fl32xx_priv) +
> >>> +		      (sizeof(struct is31fl32xx_led_data) * num_leds);
> >>> +}
> >>> +
> >>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
> >>> +				     const struct device_node *child,
> >>> +				     struct is31fl32xx_led_data *led_data)
> >>> +{
> >>> +	struct led_classdev *cdev = &led_data->cdev;
> >>> +	int ret = 0;
> >>> +	u32 reg;
> >>> +
> >>> +	if (of_property_read_string(child, "label", &cdev->name))
> >>> +		cdev->name = child->name;
> >>> +
> >>> +	ret = of_property_read_u32(child, "reg", &reg);
> >>> +	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
> >>> +		dev_err(dev,
> >>> +			"Child node %s does not have a valid reg property\n",
> >>> +			child->full_name);
> >>> +		return -EINVAL;
> >>> +	}
> >>> +	led_data->channel = reg;
> >>> +
> >>> +	of_property_read_string(child, "linux,default-trigger",
> >>> +				&cdev->default_trigger);
> >>> +
> >>> +	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
> >>> +
> >>> +	return 0;
> >>> +}
> >>> +
> >>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
> >>> +					struct is31fl32xx_priv *priv,
> >>> +					u8 channel)
> >>> +{
> >>> +	size_t i;
> >>> +
> >>> +	for (i = 0; i < priv->num_leds; i++) {
> >>> +		if (priv->leds[i].channel == channel)
> >>> +			return &priv->leds[i];
> >>> +	}
> >>> +
> >>> +	return NULL;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_parse_dt(struct device *dev,
> >>> +			       struct is31fl32xx_priv *priv)
> >>> +{
> >>> +	struct device_node *child;
> >>> +	int ret = 0;
> >>> +
> >>> +	for_each_child_of_node(dev->of_node, child) {
> >>> +		struct is31fl32xx_led_data *led_data =
> >>> +			&priv->leds[priv->num_leds];
> >>> +		const struct is31fl32xx_led_data *other_led_data;
> >>> +
> >>> +		led_data->priv = priv;
> >>> +
> >>> +		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
> >>> +		if (ret)
> >>> +			goto err;
> >>> +
> >>> +		/* Detect if channel is already in use by another child */
> >>> +		other_led_data = is31fl32xx_find_led_data(priv,
> >>> +							  led_data->channel);
> >>> +		if (other_led_data) {
> >>> +			dev_err(dev,
> >>> +				"%s and %s both attempting to use channel %d\n",
> >>> +				led_data->cdev.name,
> >>> +				other_led_data->cdev.name,
> >>> +				led_data->channel);
> >>> +			goto err;
> >>> +		}
> >>> +
> >>> +		ret = devm_led_classdev_register(dev, &led_data->cdev);
> >>> +		if (ret) {
> >>> +			dev_err(dev, "failed to register PWM led for %s: %d\n",
> >>> +				led_data->cdev.name, ret);
> >>> +			goto err;
> >>> +		}
> >>> +
> >>> +		priv->num_leds++;
> >>> +	}
> >>> +
> >>> +	return 0;
> >>> +
> >>> +err:
> >>> +	of_node_put(child);
> >>> +	return ret;
> >>> +}
> >>> +
> >>> +static const struct of_device_id of_is31fl31xx_match[] = {
> >>> +	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
> >>> +	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
> >>> +	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
> >>> +	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
> >>> +	{},
> >>> +};
> >>> +
> >>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
> >>> +
> >>> +static int is31fl32xx_probe(struct i2c_client *client,
> >>> +			    const struct i2c_device_id *id)
> >>> +{
> >>> +	const struct is31fl32xx_chipdef *cdef;
> >>> +	const struct of_device_id *of_dev_id;
> >>> +	struct device *dev = &client->dev;
> >>> +	struct is31fl32xx_priv *priv;
> >>> +	int count;
> >>> +	int ret = 0;
> >>> +
> >>> +	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
> >>> +	if (!of_dev_id)
> >>> +		return -EINVAL;
> >>> +
> >>> +	cdef = of_dev_id->data;
> >>> +
> >>> +	count = of_get_child_count(dev->of_node);
> >>> +	if (!count)
> >>> +		return -EINVAL;
> >>> +
> >>> +	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
> >>> +			    GFP_KERNEL);
> >>> +	if (!priv)
> >>> +		return -ENOMEM;
> >>> +
> >>> +	priv->client = client;
> >>> +	priv->cdef = cdef;
> >>> +	i2c_set_clientdata(client, priv);
> >>> +
> >>> +	ret = is31fl32xx_init_regs(priv);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	ret = is31fl32xx_parse_dt(dev, priv);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	return 0;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_remove(struct i2c_client *client)
> >>> +{
> >>> +	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
> >>> +
> >>> +	return is31fl32xx_reset_regs(priv);
> >>> +}
> >>> +
> >>> +/*
> >>> + * i2c-core requires that id_table be non-NULL, even though
> >>> + * it is not used for DeviceTree based instantiation.
> >>> + */
> >>> +static const struct i2c_device_id is31fl31xx_id[] = {
> >>> +	{},
> >>> +};
> >>> +
> >>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
> >>> +
> >>> +static struct i2c_driver is31fl32xx_driver = {
> >>> +	.driver = {
> >>> +		.name	= "is31fl32xx",
> >>> +		.of_match_table = of_is31fl31xx_match,
> >>> +	},
> >>> +	.probe		= is31fl32xx_probe,
> >>> +	.remove		= is31fl32xx_remove,
> >>> +	.id_table	= is31fl31xx_id,
> >>> +};
> >>> +
> >>> +module_i2c_driver(is31fl32xx_driver);
> >>> +
> >>> +MODULE_AUTHOR("David Rivshin <drivshin-5fOYsn7Fw8lBDgjK7y7TUQ@public.gmane.org>");
> >>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
> >>> +MODULE_LICENSE("GPL v2");
> >>>  

--
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 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[David Rivshin (Allworx)]

On Fri, 04 Mar 2016 08:54:02 +0100
Jacek Anaszewski <j.anaszewski@samsung.com> wrote:

> On 03/04/2016 01:45 AM, David Rivshin (Allworx) wrote:
> > On Thu, 03 Mar 2016 15:51:32 +0100
> > Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
> >  
> >> Hi David,
> >>
> >> Thanks for the update. Two remarks in the code.
> >>
> >> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:  
> >>> From: David Rivshin <drivshin@allworx.com>
> >>>
> >>> The IS31FL32xx family of LED controllers are I2C devices with multiple
> >>> constant-current channels, each with independent 256-level PWM control.
> >>>
> >>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >>>
> >>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
> >>> (TI am335x) platform.
> >>>
> >>> The programming paradigm of these devices is similar in the following
> >>> ways:
> >>>    - All registers are 8 bit
> >>>    - All LED control registers are write-only
> >>>    - Each LED channel has a PWM register (0-255)
> >>>    - PWM register writes are shadowed until an Update register is poked
> >>>    - All have a concept of Software Shutdown, which disables output
> >>>
> >>> However, there are some differences in devices:
> >>>    - 3236/3235 have a separate Control register for each LED,
> >>>      (3218/3216 pack the enable bits into fewer registers)
> >>>    - 3236/3235 have a per-channel current divisor setting
> >>>    - 3236/3235 have a Global Control register that can turn off all LEDs
> >>>    - 3216 is unique in a number of ways
> >>>       - OUT9-OUT16 can be configured as GPIOs instead of LED controls
> >>>       - LEDs can be programmed with an 8-frame animation, with
> >>>         programmable delay between frames
> >>>       - LEDs can be modulated by an input audio signal
> >>>       - Max output current can be adjusted from 1/4 to 2x globally
> >>>       - Has a Configuration register instead of a Shutdown register
> >>>
> >>> This driver currently only supports the base PWM control function
> >>> of these devices. The following features of these devices are not
> >>> implemented, although it should be possible to add them in the future:
> >>>    - All devices are capable of going into a lower-power "software
> >>>      shutdown" mode.
> >>>    - The is31fl3236 and is31fl3235 can reduce the max output current
> >>>      per-channel with a divisor of 1, 2, 3, or 4.
> >>>    - The is31fl3216 can use some LED channels as GPIOs instead.
> >>>    - The is31fl3216 can animate LEDs in hardware.
> >>>    - The is31fl3216 can modulate LEDs according to an audio input.
> >>>    - The is31fl3216 can reduce/increase max output current globally.
> >>>
> >>> Signed-off-by: David Rivshin <drivshin@allworx.com>
> >>> ---
> >>>
> >>> You may see two instances of this warning:
> >>>     "passing argument 1 of 'of_property_read_string' discards 'const'
> >>>      qualifier from pointer target type"
> >>> That is a result of of_property_read_string() taking a non-const
> >>> struct device_node pointer parameter. I have separately submitted a
> >>> patch to fix that [1], and a few related functions which had the same
> >>> issue. I'm hoping that will get into linux-next before this does, so
> >>> that the warnings never show up there.  
> >>
> >> Please adjust the patch so that it compiles without warnings on
> >> current linux-next. Your patch for DT API hasn't been reviewed yet
> >> AFICS, and I can imagine that there will be some resistance against.  
> >
> > Since the DT API patch was just accepted by Rob [1], would it be OK
> > to wait for the results of Stefan's testing (and any other reviews)
> > before making a decision on this? From Stefan's note, it won't be
> > until this weekend that he will have a chance to test, and I'm
> > guessing the DT API patch will make its way through Rob's tree to
> > linux-next by then.  
> 
> OK.
> 
> > FYI, the warning workaround would be to make the second parameter to
> > is31fl32xx_parse_child_dt() non-const.
> >
> > [1] https://lkml.org/lkml/2016/3/3/924
> >  
> >>> Changes from RFC:
> >>>    - Removed max-brightness DT property.
> >>>    - Refer to these devices as "LED controllers" in Kconfig.
> >>>    - Removed redundant last sentence from Kconfig entry
> >>>    - Removed unnecessary debug code.
> >>>    - Do not set led_classdev.brightness to 0 explicitly, as it is
> >>>      already initialized to 0 by devm_kzalloc().
> >>>    - Used of_property_read_string() instead of of_get_property().
> >>>    - Fail immediately on DT parsing error in a child node, rather than
> >>>      continuing on with the non-faulty ones.
> >>>    - Added additional comments for some things that might be non-obvious.
> >>>    - Added constants for the location of the SSD bit in the SHUTDOWN
> >>>      register, and the 3216's CONFIG register.
> >>>    - Added special sw_shutdown_func for the 3216 device, as that bit
> >>>      is in a different register, at a different position, and has reverse
> >>>      polarity compared to all the other devices.
> >>>    - Refactored is31fl32xx_init_regs() to separate out some logic into
> >>>      is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
> >>>
> >>> [1] https://lkml.org/lkml/2016/3/2/746
> >>>
> >>>    drivers/leds/Kconfig           |   8 +
> >>>    drivers/leds/Makefile          |   1 +
> >>>    drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
> >>>    3 files changed, 514 insertions(+)
> >>>    create mode 100644 drivers/leds/leds-is31fl32xx.c
> >>>
> >>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> >>> index 1034696..9c63ba4 100644
> >>> --- a/drivers/leds/Kconfig
> >>> +++ b/drivers/leds/Kconfig
> >>> @@ -580,6 +580,14 @@ config LEDS_SN3218
> >>>    	  This driver can also be built as a module. If so the module
> >>>    	  will be called leds-sn3218.
> >>>
> >>> +config LEDS_IS31FL32XX
> >>> +	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
> >>> +	depends on LEDS_CLASS && I2C && OF
> >>> +	help
> >>> +	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
> >>> +	  They are I2C devices with multiple constant-current channels, each
> >>> +	  with independent 256-level PWM control.
> >>> +
> >>>    comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
> >>>
> >>>    config LEDS_BLINKM
> >>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
> >>> index 89c9b6f..3fdf313 100644
> >>> --- a/drivers/leds/Makefile
> >>> +++ b/drivers/leds/Makefile
> >>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
> >>>    obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
> >>>    obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
> >>>    obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
> >>> +obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
> >>>
> >>>    # LED SPI Drivers
> >>>    obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
> >>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
> >>> new file mode 100644
> >>> index 0000000..49818f0
> >>> --- /dev/null
> >>> +++ b/drivers/leds/leds-is31fl32xx.c
> >>> @@ -0,0 +1,505 @@
> >>> +/*
> >>> + * linux/drivers/leds-is31fl32xx.c
> >>> + *
> >>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
> >>> + *
> >>> + * Copyright 2015 Allworx Corp.
> >>> + *
> >>> + *
> >>> + * 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.
> >>> + *
> >>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >>> + */
> >>> +
> >>> +#include <linux/err.h>
> >>> +#include <linux/i2c.h>
> >>> +#include <linux/kernel.h>
> >>> +#include <linux/leds.h>
> >>> +#include <linux/module.h>
> >>> +#include <linux/of_platform.h>
> >>> +
> >>> +/* Used to indicate a device has no such register */
> >>> +#define IS31FL32XX_REG_NONE 0xFF
> >>> +
> >>> +/* Software Shutdown bit in Shutdown Register */
> >>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
> >>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
> >>> +
> >>> +/* IS31FL3216 has a number of unique registers */
> >>> +#define IS31FL3216_CONFIG_REG 0x00
> >>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
> >>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
> >>> +
> >>> +/* Software Shutdown bit in 3216 Config Register */
> >>> +#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
> >>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
> >>> +
> >>> +struct is31fl32xx_priv;
> >>> +struct is31fl32xx_led_data {
> >>> +	struct led_classdev cdev;
> >>> +	u8 channel; /* 1-based, max priv->cdef->channels */
> >>> +	struct is31fl32xx_priv *priv;
> >>> +};
> >>> +
> >>> +struct is31fl32xx_priv {
> >>> +	const struct is31fl32xx_chipdef *cdef;
> >>> +	struct i2c_client *client;
> >>> +	unsigned int num_leds;
> >>> +	struct is31fl32xx_led_data leds[0];
> >>> +};
> >>> +
> >>> +/**
> >>> + * struct is31fl32xx_chipdef - chip-specific attributes
> >>> + * @channels            : Number of LED channels
> >>> + * @shutdown_reg        : address of Shutdown register (optional)
> >>> + * @pwm_update_reg      : address of PWM Update register
> >>> + * @global_control_reg  : address of Global Control register (optional)
> >>> + * @reset_reg           : address of Reset register (optional)
> >>> + * @pwm_register_base   : address of first PWM register
> >>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
> >>> + * @led_control_register_base : address of first LED control register (optional)
> >>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
> >>> + * @reset_func:         : pointer to reset function
> >>> + *
> >>> + * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
> >>> + * indicates that this chip has no such register.
> >>> + *
> >>> + * If non-NULL, @reset_func will be called during probing to set all
> >>> + * necessary registers to a known initialization state. This is needed
> >>> + * for chips that do not have a @reset_reg.
> >>> + *
> >>> + * @enable_bits_per_led_control_register must be >=1 if
> >>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
> >>> + */
> >>> +struct is31fl32xx_chipdef {
> >>> +	u8	channels;
> >>> +	u8	shutdown_reg;
> >>> +	u8	pwm_update_reg;
> >>> +	u8	global_control_reg;
> >>> +	u8	reset_reg;
> >>> +	u8	pwm_register_base;
> >>> +	bool	pwm_registers_reversed;
> >>> +	u8	led_control_register_base;
> >>> +	u8	enable_bits_per_led_control_register;
> >>> +	int (*reset_func)(struct is31fl32xx_priv *priv);
> >>> +	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
> >>> +};
> >>> +
> >>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
> >>> +	.channels				= 36,
> >>> +	.shutdown_reg				= 0x00,
> >>> +	.pwm_update_reg				= 0x25,
> >>> +	.global_control_reg			= 0x4a,
> >>> +	.reset_reg				= 0x4f,
> >>> +	.pwm_register_base			= 0x01,
> >>> +	.led_control_register_base		= 0x26,
> >>> +	.enable_bits_per_led_control_register	= 1,
> >>> +};
> >>> +
> >>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
> >>> +	.channels				= 28,
> >>> +	.shutdown_reg				= 0x00,
> >>> +	.pwm_update_reg				= 0x25,
> >>> +	.global_control_reg			= 0x4a,
> >>> +	.reset_reg				= 0x4f,
> >>> +	.pwm_register_base			= 0x05,
> >>> +	.led_control_register_base		= 0x2a,
> >>> +	.enable_bits_per_led_control_register	= 1,
> >>> +};
> >>> +
> >>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
> >>> +	.channels				= 18,
> >>> +	.shutdown_reg				= 0x00,
> >>> +	.pwm_update_reg				= 0x16,
> >>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
> >>> +	.reset_reg				= 0x17,
> >>> +	.pwm_register_base			= 0x01,
> >>> +	.led_control_register_base		= 0x13,
> >>> +	.enable_bits_per_led_control_register	= 6,
> >>> +};
> >>> +
> >>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
> >>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> >>> +					bool enable);
> >>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
> >>> +	.channels				= 16,
> >>> +	.shutdown_reg				= IS31FL32XX_REG_NONE,
> >>> +	.pwm_update_reg				= 0xB0,
> >>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
> >>> +	.reset_reg				= IS31FL32XX_REG_NONE,
> >>> +	.pwm_register_base			= 0x10,
> >>> +	.pwm_registers_reversed			= true,
> >>> +	.led_control_register_base		= 0x01,
> >>> +	.enable_bits_per_led_control_register	= 8,
> >>> +	.reset_func				= is31fl3216_reset,
> >>> +	.sw_shutdown_func			= is31fl3216_software_shutdown,
> >>> +};
> >>> +
> >>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
> >>> +{
> >>> +	int ret;
> >>> +
> >>> +	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
> >>> +
> >>> +	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
> >>> +	if (ret) {
> >>> +		dev_err(&priv->client->dev,
> >>> +			"register write to 0x%02X failed (error %d)",
> >>> +			reg, ret);
> >>> +	}
> >>> +	return ret;
> >>> +}
> >>> +
> >>> +/*
> >>> + * Custom reset function for IS31FL3216 because it does not have a RESET
> >>> + * register the way that the other IS31FL32xx chips do. We don't bother
> >>> + * writing the GPIO and animation registers, because the registers we
> >>> + * do write ensure those will have no effect.
> >>> + */
> >>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
> >>> +{
> >>> +	unsigned int i;
> >>> +	int ret;
> >>> +
> >>> +	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
> >>> +			       IS31FL3216_CONFIG_SSD_ENABLE);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +	for (i = 0; i < priv->cdef->channels; i++) {
> >>> +		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
> >>> +				       0x00);
> >>> +		if (ret)
> >>> +			return ret;
> >>> +	}
> >>> +	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	return 0;
> >>> +}
> >>> +
> >>> +/*
> >>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
> >>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
> >>> + * We don't bother doing a read/modify/write on the CONFIG register because
> >>> + * we only ever use a value of '0' for the other fields in that register.
> >>> + */
> >>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> >>> +					bool enable)
> >>> +{
> >>> +	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
> >>> +			    IS31FL3216_CONFIG_SSD_DISABLE;
> >>> +
> >>> +	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
> >>> +}
> >>> +
> >>> +/*
> >>> + * NOTE: A mutex is not needed in this function because:
> >>> + * - All referenced data is read-only after probe()
> >>> + * - The I2C core has a mutex on to protect the bus
> >>> + * - There are no read/modify/write operations
> >>> + * - Intervening operations between the write of the PWM register
> >>> + *   and the Update register are harmless.
> >>> + *
> >>> + * Example:
> >>> + *	PWM_REG_1 write 16
> >>> + *	UPDATE_REG write 0
> >>> + *	PWM_REG_2 write 128
> >>> + *	UPDATE_REG write 0
> >>> + *   vs:
> >>> + *	PWM_REG_1 write 16
> >>> + *	PWM_REG_2 write 128
> >>> + *	UPDATE_REG write 0
> >>> + *	UPDATE_REG write 0
> >>> + * are equivalent. Poking the Update register merely applies all PWM
> >>> + * register writes up to that point.
> >>> + */
> >>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
> >>> +				     enum led_brightness brightness)
> >>> +{
> >>> +	const struct is31fl32xx_led_data *led_data =
> >>> +		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
> >>> +	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
> >>> +	u8 pwm_register_offset;
> >>> +	int ret;
> >>> +
> >>> +	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
> >>> +
> >>> +	/* NOTE: led_data->channel is 1-based */
> >>> +	if (cdef->pwm_registers_reversed)
> >>> +		pwm_register_offset = cdef->channels - led_data->channel;
> >>> +	else
> >>> +		pwm_register_offset = led_data->channel - 1;
> >>> +
> >>> +	ret = is31fl32xx_write(led_data->priv,
> >>> +			       cdef->pwm_register_base + pwm_register_offset,
> >>> +			       brightness);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
> >>> +}
> >>> +
> >>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
> >>> +{
> >>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>> +	int ret;
> >>> +
> >>> +	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
> >>> +		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
> >>> +		if (ret)
> >>> +			return ret;
> >>> +	}
> >>> +
> >>> +	if (cdef->reset_func)
> >>> +		return cdef->reset_func(priv);
> >>> +
> >>> +	return 0;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
> >>> +					bool enable)
> >>> +{
> >>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>> +	int ret;
> >>> +
> >>> +	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
> >>> +		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
> >>> +				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
> >>> +		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
> >>> +		if (ret)
> >>> +			return ret;
> >>> +	}
> >>> +
> >>> +	if (cdef->sw_shutdown_func)
> >>> +		return cdef->sw_shutdown_func(priv, enable);  
> >>
> >> You seem to call sw_shutdown_func only here, so why should we have
> >> enable parameter in this op?  
> >
> > I'm not sure if I understand the question, but I will try to answer.
> >
> > 'enable' is passed through is31fl32xx_software_shutdown to
> > cdef->sw_shutdown_func, so it can be either true or false at that
> > point. The purpose of sw_shutdown_func is to add any special behavior
> > when enabling/disabling software-shutdown mode, which is needed for
> > the 3216 because its SSD bit is in a different position and with
> > opposite polarity.
> >
> > Is it that 'enable' in that line of code makes it look like it's being
> > called with an hardcoded value rather than a variable? If so, perhaps a
> > different parameter name would make it more obvious? Or a kerneldoc
> > comment for the function to describe the parameter?
> >
> > Or have I totally missed the point of the question?  
> 
> Actually I should have placed this question next to
> the call to s31fl32xx_software_shutdown() in is31fl32xx_init_regs(),
> which is passed "false" in the second argument, and there is no
> other call to s31fl32xx_software_shutdown() in the driver.
> 
> Having the argument makes people wondering that there is some
> use case in the driver, where "true" is passed, but it seems not
> to be the case.

Thanks for the clarification.

Yes, there is currently no explicit call to enable software-shutdown
mode. Since the reset state of these devices is to have software-shutdown 
enabled, the only explicit operation on it that's currently needed is
to disable it.

Reasons I can think to have the 'enable' parameter anyways include:
- It seems logical to me that an API to manipulate the software-shutdown
  state should allow both enabling and disabling. 
- Having a parameter for that seemed the most obvious way to go, and it
  was trivial to implement. 
- Alternatively having separate "enable" and "disable" functions would 
  duplicate most of the logic, vs the parameter being just a single 
  conditional. And that would also imply two function pointers in the 
  chipdefs, which I'd prefer to minimize.
- If anyone wanted to implement suspend/resume in the future, they would
  most likely do it by enabling/disabling software-shutdown. Supporting
  both enable/disable from the start should make that trivial to do.
- I thought the code read better with a bool parameter, vs a longer 
  function name.

So nothing really critical, but mostly just my aesthetic and preference.

Also, I expect that is31fl32xx_software_shutdown() would be inlined, so 
the conditional check in there is optimized out anyways, and there is no 
performance penalty. Looking at the disassembly in my ARMv7a build, both 
of those have indeed happened there.

> >>> +	return 0;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
> >>> +{
> >>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>> +	int ret;
> >>> +
> >>> +	ret = is31fl32xx_reset_regs(priv);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	/*
> >>> +	 * Set enable bit for all channels.
> >>> +	 * We will control state with PWM registers alone.
> >>> +	 */
> >>> +	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
> >>> +		u8 value =
> >>> +		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
> >>> +		u8 num_regs = cdef->channels /
> >>> +				cdef->enable_bits_per_led_control_register;
> >>> +		int i;
> >>> +
> >>> +		for (i = 0; i < num_regs; i++) {
> >>> +			ret = is31fl32xx_write(priv,
> >>> +					       cdef->led_control_register_base+i,
> >>> +					       value);
> >>> +			if (ret)
> >>> +				return ret;
> >>> +		}
> >>> +	}
> >>> +
> >>> +	ret = is31fl32xx_software_shutdown(priv, false);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
> >>> +		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
> >>> +		if (ret)
> >>> +			return ret;
> >>> +	}
> >>> +
> >>> +	return 0;
> >>> +}
> >>> +
> >>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
> >>> +{
> >>> +	return sizeof(struct is31fl32xx_priv) +
> >>> +		      (sizeof(struct is31fl32xx_led_data) * num_leds);
> >>> +}
> >>> +
> >>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
> >>> +				     const struct device_node *child,
> >>> +				     struct is31fl32xx_led_data *led_data)
> >>> +{
> >>> +	struct led_classdev *cdev = &led_data->cdev;
> >>> +	int ret = 0;
> >>> +	u32 reg;
> >>> +
> >>> +	if (of_property_read_string(child, "label", &cdev->name))
> >>> +		cdev->name = child->name;
> >>> +
> >>> +	ret = of_property_read_u32(child, "reg", &reg);
> >>> +	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
> >>> +		dev_err(dev,
> >>> +			"Child node %s does not have a valid reg property\n",
> >>> +			child->full_name);
> >>> +		return -EINVAL;
> >>> +	}
> >>> +	led_data->channel = reg;
> >>> +
> >>> +	of_property_read_string(child, "linux,default-trigger",
> >>> +				&cdev->default_trigger);
> >>> +
> >>> +	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
> >>> +
> >>> +	return 0;
> >>> +}
> >>> +
> >>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
> >>> +					struct is31fl32xx_priv *priv,
> >>> +					u8 channel)
> >>> +{
> >>> +	size_t i;
> >>> +
> >>> +	for (i = 0; i < priv->num_leds; i++) {
> >>> +		if (priv->leds[i].channel == channel)
> >>> +			return &priv->leds[i];
> >>> +	}
> >>> +
> >>> +	return NULL;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_parse_dt(struct device *dev,
> >>> +			       struct is31fl32xx_priv *priv)
> >>> +{
> >>> +	struct device_node *child;
> >>> +	int ret = 0;
> >>> +
> >>> +	for_each_child_of_node(dev->of_node, child) {
> >>> +		struct is31fl32xx_led_data *led_data =
> >>> +			&priv->leds[priv->num_leds];
> >>> +		const struct is31fl32xx_led_data *other_led_data;
> >>> +
> >>> +		led_data->priv = priv;
> >>> +
> >>> +		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
> >>> +		if (ret)
> >>> +			goto err;
> >>> +
> >>> +		/* Detect if channel is already in use by another child */
> >>> +		other_led_data = is31fl32xx_find_led_data(priv,
> >>> +							  led_data->channel);
> >>> +		if (other_led_data) {
> >>> +			dev_err(dev,
> >>> +				"%s and %s both attempting to use channel %d\n",
> >>> +				led_data->cdev.name,
> >>> +				other_led_data->cdev.name,
> >>> +				led_data->channel);
> >>> +			goto err;
> >>> +		}
> >>> +
> >>> +		ret = devm_led_classdev_register(dev, &led_data->cdev);
> >>> +		if (ret) {
> >>> +			dev_err(dev, "failed to register PWM led for %s: %d\n",
> >>> +				led_data->cdev.name, ret);
> >>> +			goto err;
> >>> +		}
> >>> +
> >>> +		priv->num_leds++;
> >>> +	}
> >>> +
> >>> +	return 0;
> >>> +
> >>> +err:
> >>> +	of_node_put(child);
> >>> +	return ret;
> >>> +}
> >>> +
> >>> +static const struct of_device_id of_is31fl31xx_match[] = {
> >>> +	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
> >>> +	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
> >>> +	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
> >>> +	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
> >>> +	{},
> >>> +};
> >>> +
> >>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
> >>> +
> >>> +static int is31fl32xx_probe(struct i2c_client *client,
> >>> +			    const struct i2c_device_id *id)
> >>> +{
> >>> +	const struct is31fl32xx_chipdef *cdef;
> >>> +	const struct of_device_id *of_dev_id;
> >>> +	struct device *dev = &client->dev;
> >>> +	struct is31fl32xx_priv *priv;
> >>> +	int count;
> >>> +	int ret = 0;
> >>> +
> >>> +	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
> >>> +	if (!of_dev_id)
> >>> +		return -EINVAL;
> >>> +
> >>> +	cdef = of_dev_id->data;
> >>> +
> >>> +	count = of_get_child_count(dev->of_node);
> >>> +	if (!count)
> >>> +		return -EINVAL;
> >>> +
> >>> +	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
> >>> +			    GFP_KERNEL);
> >>> +	if (!priv)
> >>> +		return -ENOMEM;
> >>> +
> >>> +	priv->client = client;
> >>> +	priv->cdef = cdef;
> >>> +	i2c_set_clientdata(client, priv);
> >>> +
> >>> +	ret = is31fl32xx_init_regs(priv);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	ret = is31fl32xx_parse_dt(dev, priv);
> >>> +	if (ret)
> >>> +		return ret;
> >>> +
> >>> +	return 0;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_remove(struct i2c_client *client)
> >>> +{
> >>> +	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
> >>> +
> >>> +	return is31fl32xx_reset_regs(priv);
> >>> +}
> >>> +
> >>> +/*
> >>> + * i2c-core requires that id_table be non-NULL, even though
> >>> + * it is not used for DeviceTree based instantiation.
> >>> + */
> >>> +static const struct i2c_device_id is31fl31xx_id[] = {
> >>> +	{},
> >>> +};
> >>> +
> >>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
> >>> +
> >>> +static struct i2c_driver is31fl32xx_driver = {
> >>> +	.driver = {
> >>> +		.name	= "is31fl32xx",
> >>> +		.of_match_table = of_is31fl31xx_match,
> >>> +	},
> >>> +	.probe		= is31fl32xx_probe,
> >>> +	.remove		= is31fl32xx_remove,
> >>> +	.id_table	= is31fl31xx_id,
> >>> +};
> >>> +
> >>> +module_i2c_driver(is31fl32xx_driver);
> >>> +
> >>> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
> >>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
> >>> +MODULE_LICENSE("GPL v2");
> >>>  

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[Jacek Anaszewski]

On 03/04/2016 03:27 PM, David Rivshin (Allworx) wrote:
> (Stefan, sorry for the duplicate, I just realized that I originally
> replied only to you by accident).
>
> On Thu, 3 Mar 2016 19:13:03 +0100 (CET)
> Stefan Wahren <stefan.wahren@i2se.com> wrote:
>
>> Hi David,
>>
>> i will test the driver on weekend. Some comments below
>
> Great, thanks very much.
>
>>> "David Rivshin (Allworx)" <drivshin.allworx@gmail.com> hat am 3. März 2016 um
>>> 04:01 geschrieben:
>>>
>>>
>>> From: David Rivshin <drivshin@allworx.com>
>>>
>>> The IS31FL32xx family of LED controllers are I2C devices with multiple
>>> constant-current channels, each with independent 256-level PWM control.
>>>
>>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>>
>>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
>>> (TI am335x) platform.
>>>
>>> The programming paradigm of these devices is similar in the following
>>> ways:
>>> - All registers are 8 bit
>>> - All LED control registers are write-only
>>> - Each LED channel has a PWM register (0-255)
>>> - PWM register writes are shadowed until an Update register is poked
>>> - All have a concept of Software Shutdown, which disables output
>>>
>>> However, there are some differences in devices:
>>> - 3236/3235 have a separate Control register for each LED,
>>> (3218/3216 pack the enable bits into fewer registers)
>>> - 3236/3235 have a per-channel current divisor setting
>>> - 3236/3235 have a Global Control register that can turn off all LEDs
>>> - 3216 is unique in a number of ways
>>> - OUT9-OUT16 can be configured as GPIOs instead of LED controls
>>> - LEDs can be programmed with an 8-frame animation, with
>>> programmable delay between frames
>>> - LEDs can be modulated by an input audio signal
>>> - Max output current can be adjusted from 1/4 to 2x globally
>>> - Has a Configuration register instead of a Shutdown register
>>>
>>> This driver currently only supports the base PWM control function
>>> of these devices. The following features of these devices are not
>>> implemented, although it should be possible to add them in the future:
>>> - All devices are capable of going into a lower-power "software
>>> shutdown" mode.
>>> - The is31fl3236 and is31fl3235 can reduce the max output current
>>> per-channel with a divisor of 1, 2, 3, or 4.
>>> - The is31fl3216 can use some LED channels as GPIOs instead.
>>> - The is31fl3216 can animate LEDs in hardware.
>>> - The is31fl3216 can modulate LEDs according to an audio input.
>>> - The is31fl3216 can reduce/increase max output current globally.
>>>
>>> Signed-off-by: David Rivshin <drivshin@allworx.com>
>>> ---
>>>
>>> You may see two instances of this warning:
>>> "passing argument 1 of 'of_property_read_string' discards 'const'
>>> qualifier from pointer target type"
>>> That is a result of of_property_read_string() taking a non-const
>>> struct device_node pointer parameter. I have separately submitted a
>>> patch to fix that [1], and a few related functions which had the same
>>> issue. I'm hoping that will get into linux-next before this does, so
>>> that the warnings never show up there.
>>>
>>> Changes from RFC:
>>> - Removed max-brightness DT property.
>>> - Refer to these devices as "LED controllers" in Kconfig.
>>> - Removed redundant last sentence from Kconfig entry
>>> - Removed unnecessary debug code.
>>> - Do not set led_classdev.brightness to 0 explicitly, as it is
>>> already initialized to 0 by devm_kzalloc().
>>> - Used of_property_read_string() instead of of_get_property().
>>> - Fail immediately on DT parsing error in a child node, rather than
>>> continuing on with the non-faulty ones.
>>> - Added additional comments for some things that might be non-obvious.
>>> - Added constants for the location of the SSD bit in the SHUTDOWN
>>> register, and the 3216's CONFIG register.
>>> - Added special sw_shutdown_func for the 3216 device, as that bit
>>> is in a different register, at a different position, and has reverse
>>> polarity compared to all the other devices.
>>> - Refactored is31fl32xx_init_regs() to separate out some logic into
>>> is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
>>>
>>> [1] https://lkml.org/lkml/2016/3/2/746
>>>
>>> drivers/leds/Kconfig | 8 +
>>> drivers/leds/Makefile | 1 +
>>> drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
>>> 3 files changed, 514 insertions(+)
>>> create mode 100644 drivers/leds/leds-is31fl32xx.c
>>>
>>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
>>> index 1034696..9c63ba4 100644
>>> --- a/drivers/leds/Kconfig
>>> +++ b/drivers/leds/Kconfig
>>> @@ -580,6 +580,14 @@ config LEDS_SN3218
>>> This driver can also be built as a module. If so the module
>>> will be called leds-sn3218.
>>>
>>> +config LEDS_IS31FL32XX
>>> + tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
>>> + depends on LEDS_CLASS && I2C && OF
>>> + help
>>> + Say Y here to include support for ISSI IS31FL32XX LED controllers.
>>> + They are I2C devices with multiple constant-current channels, each
>>> + with independent 256-level PWM control.
>>
>> Is it worth to mention the module name here?
>
> I noticed that some do and some don't. I don't mind adding it, but it
> also seemed like it would be obvious, and therefore unnecessary.
>
> Jacek, which do you prefer?

I agree - it's obvious, we can skip it.

>>> +
>>> comment "LED driver for blink(1) USB RGB LED is under Special HID drivers
>>> (HID_THINGM)"
>>>
>>> config LEDS_BLINKM
>>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
>>> index 89c9b6f..3fdf313 100644
>>> --- a/drivers/leds/Makefile
>>> +++ b/drivers/leds/Makefile
>>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692) += leds-ktd2692.o
>>> obj-$(CONFIG_LEDS_POWERNV) += leds-powernv.o
>>> obj-$(CONFIG_LEDS_SEAD3) += leds-sead3.o
>>> obj-$(CONFIG_LEDS_SN3218) += leds-sn3218.o
>>> +obj-$(CONFIG_LEDS_IS31FL32XX) += leds-is31fl32xx.o
>>>
>>> # LED SPI Drivers
>>> obj-$(CONFIG_LEDS_DAC124S085) += leds-dac124s085.o
>>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
>>> new file mode 100644
>>> index 0000000..49818f0
>>> --- /dev/null
>>> +++ b/drivers/leds/leds-is31fl32xx.c
>>> @@ -0,0 +1,505 @@
>>> +/*
>>> + * linux/drivers/leds-is31fl32xx.c
>>
>> I think this is unnecessary.
>
> I tend to agree. I think I used leds-pwm.c as a template, and that had
> such a comment. I assumed it was coding-style and kept it, but now I see
> that only a minority of led drivers have it. If I do another spin for
> any reason I'll remove it.
>
>>> + *
>>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
>>> + *
>>> + * Copyright 2015 Allworx Corp.
>>> + *
>>> + *
>>> + * 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.
>>> + *
>>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>> + */
>>> +
>>
>> Shouldn't we include <linux/device.h> here?
>
> Good catch. I was getting that via i2c.h, but since struct device is
> referenced explicitly in a few places, device.h should probably be
> included directly.

linux/device.h is included from linux/leds.h

>>> +#include <linux/err.h>
>>> +#include <linux/i2c.h>
>>> +#include <linux/kernel.h>
>>> +#include <linux/leds.h>
>>> +#include <linux/module.h>
>>> +#include <linux/of_platform.h>
>>> +
>>> +/* Used to indicate a device has no such register */
>>> +#define IS31FL32XX_REG_NONE 0xFF
>>> +
>>> +/* Software Shutdown bit in Shutdown Register */
>>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE 0
>>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
>>> +
>>> +/* IS31FL3216 has a number of unique registers */
>>> +#define IS31FL3216_CONFIG_REG 0x00
>>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
>>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
>>> +
>>> +/* Software Shutdown bit in 3216 Config Register */
>>> +#define IS31FL3216_CONFIG_SSD_ENABLE BIT(7)
>>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
>>> +
>>> +struct is31fl32xx_priv;
>>> +struct is31fl32xx_led_data {
>>> + struct led_classdev cdev;
>>> + u8 channel; /* 1-based, max priv->cdef->channels */
>>> + struct is31fl32xx_priv *priv;
>>> +};
>>> +
>>> +struct is31fl32xx_priv {
>>> + const struct is31fl32xx_chipdef *cdef;
>>> + struct i2c_client *client;
>>> + unsigned int num_leds;
>>> + struct is31fl32xx_led_data leds[0];
>>> +};
>>> +
>>> +/**
>>> + * struct is31fl32xx_chipdef - chip-specific attributes
>>> + * @channels : Number of LED channels
>>> + * @shutdown_reg : address of Shutdown register (optional)
>>> + * @pwm_update_reg : address of PWM Update register
>>> + * @global_control_reg : address of Global Control register (optional)
>>> + * @reset_reg : address of Reset register (optional)
>>> + * @pwm_register_base : address of first PWM register
>>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
>>> + * @led_control_register_base : address of first LED control register
>>> (optional)
>>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
>>> + * @reset_func: : pointer to reset function
>>> + *
>>> + * For all optional register addresses, the sentinel value
>>> %IS31FL32XX_REG_NONE
>>> + * indicates that this chip has no such register.
>>> + *
>>> + * If non-NULL, @reset_func will be called during probing to set all
>>> + * necessary registers to a known initialization state. This is needed
>>> + * for chips that do not have a @reset_reg.
>>> + *
>>> + * @enable_bits_per_led_control_register must be >=1 if
>>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
>>> + */
>>> +struct is31fl32xx_chipdef {
>>> + u8 channels;
>>> + u8 shutdown_reg;
>>> + u8 pwm_update_reg;
>>> + u8 global_control_reg;
>>> + u8 reset_reg;
>>> + u8 pwm_register_base;
>>> + bool pwm_registers_reversed;
>>> + u8 led_control_register_base;
>>> + u8 enable_bits_per_led_control_register;
>>> + int (*reset_func)(struct is31fl32xx_priv *priv);
>>> + int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
>>> +};
>>> +
>>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
>>> + .channels = 36,
>>> + .shutdown_reg = 0x00,
>>> + .pwm_update_reg = 0x25,
>>> + .global_control_reg = 0x4a,
>>> + .reset_reg = 0x4f,
>>> + .pwm_register_base = 0x01,
>>> + .led_control_register_base = 0x26,
>>> + .enable_bits_per_led_control_register = 1,
>>> +};
>>> +
>>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
>>> + .channels = 28,
>>> + .shutdown_reg = 0x00,
>>> + .pwm_update_reg = 0x25,
>>> + .global_control_reg = 0x4a,
>>> + .reset_reg = 0x4f,
>>> + .pwm_register_base = 0x05,
>>> + .led_control_register_base = 0x2a,
>>> + .enable_bits_per_led_control_register = 1,
>>> +};
>>> +
>>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
>>> + .channels = 18,
>>> + .shutdown_reg = 0x00,
>>> + .pwm_update_reg = 0x16,
>>> + .global_control_reg = IS31FL32XX_REG_NONE,
>>> + .reset_reg = 0x17,
>>> + .pwm_register_base = 0x01,
>>> + .led_control_register_base = 0x13,
>>> + .enable_bits_per_led_control_register = 6,
>>> +};
>>> +
>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>> + bool enable);
>>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
>>> + .channels = 16,
>>> + .shutdown_reg = IS31FL32XX_REG_NONE,
>>> + .pwm_update_reg = 0xB0,
>>> + .global_control_reg = IS31FL32XX_REG_NONE,
>>> + .reset_reg = IS31FL32XX_REG_NONE,
>>> + .pwm_register_base = 0x10,
>>> + .pwm_registers_reversed = true,
>>> + .led_control_register_base = 0x01,
>>> + .enable_bits_per_led_control_register = 8,
>>> + .reset_func = is31fl3216_reset,
>>> + .sw_shutdown_func = is31fl3216_software_shutdown,
>>> +};
>>> +
>>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
>>> +{
>>> + int ret;
>>> +
>>> + dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
>>> +
>>> + ret = i2c_smbus_write_byte_data(priv->client, reg, val);
>>> + if (ret) {
>>> + dev_err(&priv->client->dev,
>>> + "register write to 0x%02X failed (error %d)",
>>> + reg, ret);
>>> + }
>>
>> In case somebody use this driver as heartbeat and writing fails permanently the
>> log will be flooded.
>
> Unless I'm mistaken that would require the device/bus to fail after
> successfully probing (probe code itself bails on the first write
> failure, so there would be no flooding as a result of that). So while
> not impossible, I imagine it would be unlikely, and I'd hate to remove
> an error message for such an important condition.
>
> I suppose I could use dev_err_ratelimited() to soften any potential
> flooding, but I second guess that because:
>   - In led_core.c set_brightness_delayed() has a dev_err() that would come
>     out on each failed LED update anyways.
>   - There is precedent in other led drivers of a similar error message.
>   - Some userspace logging programs will compresses repeated messages anyways.
>
> Jacek, what is your preference on this?

Let's leave it as is. Permanent I2C bus failure is a critical error and
flooding the log would only allow to diagnose the problem quicker.

>>> + return ret;
>>> +}
>>> +
>>> +/*
>>> + * Custom reset function for IS31FL3216 because it does not have a RESET
>>> + * register the way that the other IS31FL32xx chips do. We don't bother
>>> + * writing the GPIO and animation registers, because the registers we
>>> + * do write ensure those will have no effect.
>>> + */
>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
>>> +{
>>> + unsigned int i;
>>> + int ret;
>>> +
>>> + ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
>>> + IS31FL3216_CONFIG_SSD_ENABLE);
>>> + if (ret)
>>> + return ret;
>>> + for (i = 0; i < priv->cdef->channels; i++) {
>>> + ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
>>> + 0x00);
>>> + if (ret)
>>> + return ret;
>>> + }
>>> + ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
>>> + if (ret)
>>> + return ret;
>>> + ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
>>> + if (ret)
>>> + return ret;
>>> + ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
>>> + if (ret)
>>> + return ret;
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +/*
>>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
>>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
>>> + * We don't bother doing a read/modify/write on the CONFIG register because
>>> + * we only ever use a value of '0' for the other fields in that register.
>>> + */
>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>> + bool enable)
>>> +{
>>> + u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
>>> + IS31FL3216_CONFIG_SSD_DISABLE;
>>> +
>>> + return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
>>> +}
>>> +
>>> +/*
>>> + * NOTE: A mutex is not needed in this function because:
>>> + * - All referenced data is read-only after probe()
>>> + * - The I2C core has a mutex on to protect the bus
>>> + * - There are no read/modify/write operations
>>> + * - Intervening operations between the write of the PWM register
>>> + * and the Update register are harmless.
>>> + *
>>> + * Example:
>>> + * PWM_REG_1 write 16
>>> + * UPDATE_REG write 0
>>> + * PWM_REG_2 write 128
>>> + * UPDATE_REG write 0
>>> + * vs:
>>> + * PWM_REG_1 write 16
>>> + * PWM_REG_2 write 128
>>> + * UPDATE_REG write 0
>>> + * UPDATE_REG write 0
>>> + * are equivalent. Poking the Update register merely applies all PWM
>>> + * register writes up to that point.
>>> + */
>>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
>>> + enum led_brightness brightness)
>>> +{
>>> + const struct is31fl32xx_led_data *led_data =
>>> + container_of(led_cdev, struct is31fl32xx_led_data, cdev);
>>> + const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
>>> + u8 pwm_register_offset;
>>> + int ret;
>>> +
>>> + dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
>>> +
>>> + /* NOTE: led_data->channel is 1-based */
>>> + if (cdef->pwm_registers_reversed)
>>> + pwm_register_offset = cdef->channels - led_data->channel;
>>> + else
>>> + pwm_register_offset = led_data->channel - 1;
>>> +
>>> + ret = is31fl32xx_write(led_data->priv,
>>> + cdef->pwm_register_base + pwm_register_offset,
>>> + brightness);
>>> + if (ret)
>>> + return ret;
>>> +
>>> + return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
>>> +}
>>> +
>>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
>>> +{
>>> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>> + int ret;
>>> +
>>> + if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
>>> + ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
>>> + if (ret)
>>> + return ret;
>>> + }
>>> +
>>> + if (cdef->reset_func)
>>> + return cdef->reset_func(priv);
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
>>> + bool enable)
>>> +{
>>> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>> + int ret;
>>> +
>>> + if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
>>> + u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
>>> + IS31FL32XX_SHUTDOWN_SSD_DISABLE;
>>> + ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
>>> + if (ret)
>>> + return ret;
>>> + }
>>> +
>>> + if (cdef->sw_shutdown_func)
>>> + return cdef->sw_shutdown_func(priv, enable);
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
>>> +{
>>> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>> + int ret;
>>> +
>>> + ret = is31fl32xx_reset_regs(priv);
>>> + if (ret)
>>> + return ret;
>>> +
>>> + /*
>>> + * Set enable bit for all channels.
>>> + * We will control state with PWM registers alone.
>>> + */
>>> + if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
>>> + u8 value =
>>> + GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
>>> + u8 num_regs = cdef->channels /
>>> + cdef->enable_bits_per_led_control_register;
>>> + int i;
>>> +
>>> + for (i = 0; i < num_regs; i++) {
>>> + ret = is31fl32xx_write(priv,
>>> + cdef->led_control_register_base+i,
>>> + value);
>>> + if (ret)
>>> + return ret;
>>> + }
>>> + }
>>> +
>>> + ret = is31fl32xx_software_shutdown(priv, false);
>>> + if (ret)
>>> + return ret;
>>> +
>>> + if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
>>> + ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
>>> + if (ret)
>>> + return ret;
>>> + }
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
>>> +{
>>> + return sizeof(struct is31fl32xx_priv) +
>>> + (sizeof(struct is31fl32xx_led_data) * num_leds);
>>> +}
>>> +
>>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
>>> + const struct device_node *child,
>>> + struct is31fl32xx_led_data *led_data)
>>> +{
>>> + struct led_classdev *cdev = &led_data->cdev;
>>> + int ret = 0;
>>> + u32 reg;
>>> +
>>> + if (of_property_read_string(child, "label", &cdev->name))
>>> + cdev->name = child->name;
>>> +
>>> + ret = of_property_read_u32(child, "reg", &reg);
>>> + if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
>>> + dev_err(dev,
>>> + "Child node %s does not have a valid reg property\n",
>>> + child->full_name);
>>> + return -EINVAL;
>>> + }
>>> + led_data->channel = reg;
>>> +
>>> + of_property_read_string(child, "linux,default-trigger",
>>> + &cdev->default_trigger);
>>> +
>>> + cdev->brightness_set_blocking = is31fl32xx_brightness_set;
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
>>> + struct is31fl32xx_priv *priv,
>>> + u8 channel)
>>> +{
>>> + size_t i;
>>> +
>>> + for (i = 0; i < priv->num_leds; i++) {
>>> + if (priv->leds[i].channel == channel)
>>> + return &priv->leds[i];
>>> + }
>>> +
>>> + return NULL;
>>> +}
>>> +
>>> +static int is31fl32xx_parse_dt(struct device *dev,
>>> + struct is31fl32xx_priv *priv)
>>> +{
>>> + struct device_node *child;
>>> + int ret = 0;
>>> +
>>> + for_each_child_of_node(dev->of_node, child) {
>>> + struct is31fl32xx_led_data *led_data =
>>> + &priv->leds[priv->num_leds];
>>
>> Maybe i missed something, but is it really protected against out of index
>> access?
>
> The array is allocated with size equal to the number of child nodes,
> and num_leds is incremented once for each child node parsed. So in
> order for the index to be out of bounds, the number of child nodes
> would need to increase during the probe. I assumed that the DT is
> static during probing, but if that's not the case then you're right
> that this is a potential problem. Also, this equivalent logic is
> used in leds-pwm, leds-gpio, and leds-ns2, so that gives me
> confidence that its safe.
> Unless DT overlays change that assumption?

DT overlays would matter here if child DT nodes could be dynamically
removed, i.e. if it was possible by design to dynamically unplug LEDs
from the current outputs during LED controller operation, which is not
the case for this device (and any other LED controller I am aware of).

>>> + const struct is31fl32xx_led_data *other_led_data;
>>> +
>>> + led_data->priv = priv;
>>> +
>>> + ret = is31fl32xx_parse_child_dt(dev, child, led_data);
>>> + if (ret)
>>> + goto err;
>>> +
>>> + /* Detect if channel is already in use by another child */
>>> + other_led_data = is31fl32xx_find_led_data(priv,
>>> + led_data->channel);
>>> + if (other_led_data) {
>>> + dev_err(dev,
>>> + "%s and %s both attempting to use channel %d\n",
>>> + led_data->cdev.name,
>>> + other_led_data->cdev.name,
>>> + led_data->channel);
>>> + goto err;
>>> + }
>>> +
>>> + ret = devm_led_classdev_register(dev, &led_data->cdev);
>>> + if (ret) {
>>> + dev_err(dev, "failed to register PWM led for %s: %d\n",
>>> + led_data->cdev.name, ret);
>>> + goto err;
>>> + }
>>> +
>>> + priv->num_leds++;
>>> + }
>>> +
>>> + return 0;
>>> +
>>> +err:
>>> + of_node_put(child);
>>> + return ret;
>>> +}
>>> +
>>> +static const struct of_device_id of_is31fl31xx_match[] = {
>>> + { .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
>>> + { .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
>>> + { .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
>>> + { .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
>>> + {},
>>> +};
>>> +
>>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
>>> +
>>> +static int is31fl32xx_probe(struct i2c_client *client,
>>> + const struct i2c_device_id *id)
>>> +{
>>> + const struct is31fl32xx_chipdef *cdef;
>>> + const struct of_device_id *of_dev_id;
>>> + struct device *dev = &client->dev;
>>> + struct is31fl32xx_priv *priv;
>>> + int count;
>>> + int ret = 0;
>>> +
>>> + of_dev_id = of_match_device(of_is31fl31xx_match, dev);
>>> + if (!of_dev_id)
>>> + return -EINVAL;
>>> +
>>> + cdef = of_dev_id->data;
>>> +
>>> + count = of_get_child_count(dev->of_node);
>>> + if (!count)
>>> + return -EINVAL;
>>> +
>>> + priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
>>> + GFP_KERNEL);
>>> + if (!priv)
>>> + return -ENOMEM;
>>> +
>>> + priv->client = client;
>>> + priv->cdef = cdef;
>>> + i2c_set_clientdata(client, priv);
>>> +
>>> + ret = is31fl32xx_init_regs(priv);
>>> + if (ret)
>>> + return ret;
>>> +
>>> + ret = is31fl32xx_parse_dt(dev, priv);
>>> + if (ret)
>>> + return ret;
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static int is31fl32xx_remove(struct i2c_client *client)
>>> +{
>>> + struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
>>> +
>>> + return is31fl32xx_reset_regs(priv);
>>> +}
>>> +
>>> +/*
>>> + * i2c-core requires that id_table be non-NULL, even though
>>> + * it is not used for DeviceTree based instantiation.
>>> + */
>>> +static const struct i2c_device_id is31fl31xx_id[] = {
>>> + {},
>>> +};
>>> +
>>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
>>> +
>>> +static struct i2c_driver is31fl32xx_driver = {
>>> + .driver = {
>>> + .name = "is31fl32xx",
>>> + .of_match_table = of_is31fl31xx_match,
>>> + },
>>> + .probe = is31fl32xx_probe,
>>> + .remove = is31fl32xx_remove,
>>
>> Sorry, what was the reason to skip shutdown?
>
> If I understood Jacek's last email on the topic [1] correctly, he's now
> of the opinion that the decision to turn LEDs off on reboot should be
> left to userspace, rather than done by the driver. For these devices,
> the only thing a shutdown callback would do is turn off the LEDs (through
> any of multiple methods). So, if we want to leave the state as-is on
> reboot there's no need for a shutdown callback.
>
> [1] http://www.spinics.net/lists/linux-leds/msg05644.html
>
>>> + .id_table = is31fl31xx_id,
>>> +};
>>> +
>>> +module_i2c_driver(is31fl32xx_driver);
>>> +
>>> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
>>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
>>> +MODULE_LICENSE("GPL v2");
>>> --
>>> 2.5.0
>>>
>
>
>


-- 
Best regards,
Jacek Anaszewski

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[Jacek Anaszewski]

On 03/04/2016 04:05 PM, David Rivshin (Allworx) wrote:
> On Fri, 04 Mar 2016 08:54:02 +0100
> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>
>> On 03/04/2016 01:45 AM, David Rivshin (Allworx) wrote:
>>> On Thu, 03 Mar 2016 15:51:32 +0100
>>> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>>>
>>>> Hi David,
>>>>
>>>> Thanks for the update. Two remarks in the code.
>>>>
>>>> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:
>>>>> From: David Rivshin <drivshin@allworx.com>
>>>>>
>>>>> The IS31FL32xx family of LED controllers are I2C devices with multiple
>>>>> constant-current channels, each with independent 256-level PWM control.
>>>>>
>>>>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>>>>
>>>>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
>>>>> (TI am335x) platform.
>>>>>
>>>>> The programming paradigm of these devices is similar in the following
>>>>> ways:
>>>>>     - All registers are 8 bit
>>>>>     - All LED control registers are write-only
>>>>>     - Each LED channel has a PWM register (0-255)
>>>>>     - PWM register writes are shadowed until an Update register is poked
>>>>>     - All have a concept of Software Shutdown, which disables output
>>>>>
>>>>> However, there are some differences in devices:
>>>>>     - 3236/3235 have a separate Control register for each LED,
>>>>>       (3218/3216 pack the enable bits into fewer registers)
>>>>>     - 3236/3235 have a per-channel current divisor setting
>>>>>     - 3236/3235 have a Global Control register that can turn off all LEDs
>>>>>     - 3216 is unique in a number of ways
>>>>>        - OUT9-OUT16 can be configured as GPIOs instead of LED controls
>>>>>        - LEDs can be programmed with an 8-frame animation, with
>>>>>          programmable delay between frames
>>>>>        - LEDs can be modulated by an input audio signal
>>>>>        - Max output current can be adjusted from 1/4 to 2x globally
>>>>>        - Has a Configuration register instead of a Shutdown register
>>>>>
>>>>> This driver currently only supports the base PWM control function
>>>>> of these devices. The following features of these devices are not
>>>>> implemented, although it should be possible to add them in the future:
>>>>>     - All devices are capable of going into a lower-power "software
>>>>>       shutdown" mode.
>>>>>     - The is31fl3236 and is31fl3235 can reduce the max output current
>>>>>       per-channel with a divisor of 1, 2, 3, or 4.
>>>>>     - The is31fl3216 can use some LED channels as GPIOs instead.
>>>>>     - The is31fl3216 can animate LEDs in hardware.
>>>>>     - The is31fl3216 can modulate LEDs according to an audio input.
>>>>>     - The is31fl3216 can reduce/increase max output current globally.
>>>>>
>>>>> Signed-off-by: David Rivshin <drivshin@allworx.com>
>>>>> ---
>>>>>
>>>>> You may see two instances of this warning:
>>>>>      "passing argument 1 of 'of_property_read_string' discards 'const'
>>>>>       qualifier from pointer target type"
>>>>> That is a result of of_property_read_string() taking a non-const
>>>>> struct device_node pointer parameter. I have separately submitted a
>>>>> patch to fix that [1], and a few related functions which had the same
>>>>> issue. I'm hoping that will get into linux-next before this does, so
>>>>> that the warnings never show up there.
>>>>
>>>> Please adjust the patch so that it compiles without warnings on
>>>> current linux-next. Your patch for DT API hasn't been reviewed yet
>>>> AFICS, and I can imagine that there will be some resistance against.
>>>
>>> Since the DT API patch was just accepted by Rob [1], would it be OK
>>> to wait for the results of Stefan's testing (and any other reviews)
>>> before making a decision on this? From Stefan's note, it won't be
>>> until this weekend that he will have a chance to test, and I'm
>>> guessing the DT API patch will make its way through Rob's tree to
>>> linux-next by then.
>>
>> OK.
>>
>>> FYI, the warning workaround would be to make the second parameter to
>>> is31fl32xx_parse_child_dt() non-const.
>>>
>>> [1] https://lkml.org/lkml/2016/3/3/924
>>>
>>>>> Changes from RFC:
>>>>>     - Removed max-brightness DT property.
>>>>>     - Refer to these devices as "LED controllers" in Kconfig.
>>>>>     - Removed redundant last sentence from Kconfig entry
>>>>>     - Removed unnecessary debug code.
>>>>>     - Do not set led_classdev.brightness to 0 explicitly, as it is
>>>>>       already initialized to 0 by devm_kzalloc().
>>>>>     - Used of_property_read_string() instead of of_get_property().
>>>>>     - Fail immediately on DT parsing error in a child node, rather than
>>>>>       continuing on with the non-faulty ones.
>>>>>     - Added additional comments for some things that might be non-obvious.
>>>>>     - Added constants for the location of the SSD bit in the SHUTDOWN
>>>>>       register, and the 3216's CONFIG register.
>>>>>     - Added special sw_shutdown_func for the 3216 device, as that bit
>>>>>       is in a different register, at a different position, and has reverse
>>>>>       polarity compared to all the other devices.
>>>>>     - Refactored is31fl32xx_init_regs() to separate out some logic into
>>>>>       is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
>>>>>
>>>>> [1] https://lkml.org/lkml/2016/3/2/746
>>>>>
>>>>>     drivers/leds/Kconfig           |   8 +
>>>>>     drivers/leds/Makefile          |   1 +
>>>>>     drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
>>>>>     3 files changed, 514 insertions(+)
>>>>>     create mode 100644 drivers/leds/leds-is31fl32xx.c
>>>>>
>>>>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
>>>>> index 1034696..9c63ba4 100644
>>>>> --- a/drivers/leds/Kconfig
>>>>> +++ b/drivers/leds/Kconfig
>>>>> @@ -580,6 +580,14 @@ config LEDS_SN3218
>>>>>     	  This driver can also be built as a module. If so the module
>>>>>     	  will be called leds-sn3218.
>>>>>
>>>>> +config LEDS_IS31FL32XX
>>>>> +	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
>>>>> +	depends on LEDS_CLASS && I2C && OF
>>>>> +	help
>>>>> +	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
>>>>> +	  They are I2C devices with multiple constant-current channels, each
>>>>> +	  with independent 256-level PWM control.
>>>>> +
>>>>>     comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
>>>>>
>>>>>     config LEDS_BLINKM
>>>>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
>>>>> index 89c9b6f..3fdf313 100644
>>>>> --- a/drivers/leds/Makefile
>>>>> +++ b/drivers/leds/Makefile
>>>>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
>>>>>     obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
>>>>>     obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
>>>>>     obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
>>>>> +obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
>>>>>
>>>>>     # LED SPI Drivers
>>>>>     obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
>>>>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
>>>>> new file mode 100644
>>>>> index 0000000..49818f0
>>>>> --- /dev/null
>>>>> +++ b/drivers/leds/leds-is31fl32xx.c
>>>>> @@ -0,0 +1,505 @@
>>>>> +/*
>>>>> + * linux/drivers/leds-is31fl32xx.c
>>>>> + *
>>>>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
>>>>> + *
>>>>> + * Copyright 2015 Allworx Corp.
>>>>> + *
>>>>> + *
>>>>> + * 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.
>>>>> + *
>>>>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>>>> + */
>>>>> +
>>>>> +#include <linux/err.h>
>>>>> +#include <linux/i2c.h>
>>>>> +#include <linux/kernel.h>
>>>>> +#include <linux/leds.h>
>>>>> +#include <linux/module.h>
>>>>> +#include <linux/of_platform.h>
>>>>> +
>>>>> +/* Used to indicate a device has no such register */
>>>>> +#define IS31FL32XX_REG_NONE 0xFF
>>>>> +
>>>>> +/* Software Shutdown bit in Shutdown Register */
>>>>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
>>>>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
>>>>> +
>>>>> +/* IS31FL3216 has a number of unique registers */
>>>>> +#define IS31FL3216_CONFIG_REG 0x00
>>>>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
>>>>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
>>>>> +
>>>>> +/* Software Shutdown bit in 3216 Config Register */
>>>>> +#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
>>>>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
>>>>> +
>>>>> +struct is31fl32xx_priv;
>>>>> +struct is31fl32xx_led_data {
>>>>> +	struct led_classdev cdev;
>>>>> +	u8 channel; /* 1-based, max priv->cdef->channels */
>>>>> +	struct is31fl32xx_priv *priv;
>>>>> +};
>>>>> +
>>>>> +struct is31fl32xx_priv {
>>>>> +	const struct is31fl32xx_chipdef *cdef;
>>>>> +	struct i2c_client *client;
>>>>> +	unsigned int num_leds;
>>>>> +	struct is31fl32xx_led_data leds[0];
>>>>> +};
>>>>> +
>>>>> +/**
>>>>> + * struct is31fl32xx_chipdef - chip-specific attributes
>>>>> + * @channels            : Number of LED channels
>>>>> + * @shutdown_reg        : address of Shutdown register (optional)
>>>>> + * @pwm_update_reg      : address of PWM Update register
>>>>> + * @global_control_reg  : address of Global Control register (optional)
>>>>> + * @reset_reg           : address of Reset register (optional)
>>>>> + * @pwm_register_base   : address of first PWM register
>>>>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
>>>>> + * @led_control_register_base : address of first LED control register (optional)
>>>>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
>>>>> + * @reset_func:         : pointer to reset function
>>>>> + *
>>>>> + * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
>>>>> + * indicates that this chip has no such register.
>>>>> + *
>>>>> + * If non-NULL, @reset_func will be called during probing to set all
>>>>> + * necessary registers to a known initialization state. This is needed
>>>>> + * for chips that do not have a @reset_reg.
>>>>> + *
>>>>> + * @enable_bits_per_led_control_register must be >=1 if
>>>>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
>>>>> + */
>>>>> +struct is31fl32xx_chipdef {
>>>>> +	u8	channels;
>>>>> +	u8	shutdown_reg;
>>>>> +	u8	pwm_update_reg;
>>>>> +	u8	global_control_reg;
>>>>> +	u8	reset_reg;
>>>>> +	u8	pwm_register_base;
>>>>> +	bool	pwm_registers_reversed;
>>>>> +	u8	led_control_register_base;
>>>>> +	u8	enable_bits_per_led_control_register;
>>>>> +	int (*reset_func)(struct is31fl32xx_priv *priv);
>>>>> +	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
>>>>> +};
>>>>> +
>>>>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
>>>>> +	.channels				= 36,
>>>>> +	.shutdown_reg				= 0x00,
>>>>> +	.pwm_update_reg				= 0x25,
>>>>> +	.global_control_reg			= 0x4a,
>>>>> +	.reset_reg				= 0x4f,
>>>>> +	.pwm_register_base			= 0x01,
>>>>> +	.led_control_register_base		= 0x26,
>>>>> +	.enable_bits_per_led_control_register	= 1,
>>>>> +};
>>>>> +
>>>>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
>>>>> +	.channels				= 28,
>>>>> +	.shutdown_reg				= 0x00,
>>>>> +	.pwm_update_reg				= 0x25,
>>>>> +	.global_control_reg			= 0x4a,
>>>>> +	.reset_reg				= 0x4f,
>>>>> +	.pwm_register_base			= 0x05,
>>>>> +	.led_control_register_base		= 0x2a,
>>>>> +	.enable_bits_per_led_control_register	= 1,
>>>>> +};
>>>>> +
>>>>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
>>>>> +	.channels				= 18,
>>>>> +	.shutdown_reg				= 0x00,
>>>>> +	.pwm_update_reg				= 0x16,
>>>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
>>>>> +	.reset_reg				= 0x17,
>>>>> +	.pwm_register_base			= 0x01,
>>>>> +	.led_control_register_base		= 0x13,
>>>>> +	.enable_bits_per_led_control_register	= 6,
>>>>> +};
>>>>> +
>>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
>>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>>>> +					bool enable);
>>>>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
>>>>> +	.channels				= 16,
>>>>> +	.shutdown_reg				= IS31FL32XX_REG_NONE,
>>>>> +	.pwm_update_reg				= 0xB0,
>>>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
>>>>> +	.reset_reg				= IS31FL32XX_REG_NONE,
>>>>> +	.pwm_register_base			= 0x10,
>>>>> +	.pwm_registers_reversed			= true,
>>>>> +	.led_control_register_base		= 0x01,
>>>>> +	.enable_bits_per_led_control_register	= 8,
>>>>> +	.reset_func				= is31fl3216_reset,
>>>>> +	.sw_shutdown_func			= is31fl3216_software_shutdown,
>>>>> +};
>>>>> +
>>>>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
>>>>> +{
>>>>> +	int ret;
>>>>> +
>>>>> +	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
>>>>> +
>>>>> +	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
>>>>> +	if (ret) {
>>>>> +		dev_err(&priv->client->dev,
>>>>> +			"register write to 0x%02X failed (error %d)",
>>>>> +			reg, ret);
>>>>> +	}
>>>>> +	return ret;
>>>>> +}
>>>>> +
>>>>> +/*
>>>>> + * Custom reset function for IS31FL3216 because it does not have a RESET
>>>>> + * register the way that the other IS31FL32xx chips do. We don't bother
>>>>> + * writing the GPIO and animation registers, because the registers we
>>>>> + * do write ensure those will have no effect.
>>>>> + */
>>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
>>>>> +{
>>>>> +	unsigned int i;
>>>>> +	int ret;
>>>>> +
>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
>>>>> +			       IS31FL3216_CONFIG_SSD_ENABLE);
>>>>> +	if (ret)
>>>>> +		return ret;
>>>>> +	for (i = 0; i < priv->cdef->channels; i++) {
>>>>> +		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
>>>>> +				       0x00);
>>>>> +		if (ret)
>>>>> +			return ret;
>>>>> +	}
>>>>> +	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
>>>>> +	if (ret)
>>>>> +		return ret;
>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
>>>>> +	if (ret)
>>>>> +		return ret;
>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
>>>>> +	if (ret)
>>>>> +		return ret;
>>>>> +
>>>>> +	return 0;
>>>>> +}
>>>>> +
>>>>> +/*
>>>>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
>>>>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
>>>>> + * We don't bother doing a read/modify/write on the CONFIG register because
>>>>> + * we only ever use a value of '0' for the other fields in that register.
>>>>> + */
>>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>>>> +					bool enable)
>>>>> +{
>>>>> +	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
>>>>> +			    IS31FL3216_CONFIG_SSD_DISABLE;
>>>>> +
>>>>> +	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
>>>>> +}
>>>>> +
>>>>> +/*
>>>>> + * NOTE: A mutex is not needed in this function because:
>>>>> + * - All referenced data is read-only after probe()
>>>>> + * - The I2C core has a mutex on to protect the bus
>>>>> + * - There are no read/modify/write operations
>>>>> + * - Intervening operations between the write of the PWM register
>>>>> + *   and the Update register are harmless.
>>>>> + *
>>>>> + * Example:
>>>>> + *	PWM_REG_1 write 16
>>>>> + *	UPDATE_REG write 0
>>>>> + *	PWM_REG_2 write 128
>>>>> + *	UPDATE_REG write 0
>>>>> + *   vs:
>>>>> + *	PWM_REG_1 write 16
>>>>> + *	PWM_REG_2 write 128
>>>>> + *	UPDATE_REG write 0
>>>>> + *	UPDATE_REG write 0
>>>>> + * are equivalent. Poking the Update register merely applies all PWM
>>>>> + * register writes up to that point.
>>>>> + */
>>>>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
>>>>> +				     enum led_brightness brightness)
>>>>> +{
>>>>> +	const struct is31fl32xx_led_data *led_data =
>>>>> +		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
>>>>> +	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
>>>>> +	u8 pwm_register_offset;
>>>>> +	int ret;
>>>>> +
>>>>> +	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
>>>>> +
>>>>> +	/* NOTE: led_data->channel is 1-based */
>>>>> +	if (cdef->pwm_registers_reversed)
>>>>> +		pwm_register_offset = cdef->channels - led_data->channel;
>>>>> +	else
>>>>> +		pwm_register_offset = led_data->channel - 1;
>>>>> +
>>>>> +	ret = is31fl32xx_write(led_data->priv,
>>>>> +			       cdef->pwm_register_base + pwm_register_offset,
>>>>> +			       brightness);
>>>>> +	if (ret)
>>>>> +		return ret;
>>>>> +
>>>>> +	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
>>>>> +{
>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>> +	int ret;
>>>>> +
>>>>> +	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
>>>>> +		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
>>>>> +		if (ret)
>>>>> +			return ret;
>>>>> +	}
>>>>> +
>>>>> +	if (cdef->reset_func)
>>>>> +		return cdef->reset_func(priv);
>>>>> +
>>>>> +	return 0;
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
>>>>> +					bool enable)
>>>>> +{
>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>> +	int ret;
>>>>> +
>>>>> +	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
>>>>> +		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
>>>>> +				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
>>>>> +		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
>>>>> +		if (ret)
>>>>> +			return ret;
>>>>> +	}
>>>>> +
>>>>> +	if (cdef->sw_shutdown_func)
>>>>> +		return cdef->sw_shutdown_func(priv, enable);
>>>>
>>>> You seem to call sw_shutdown_func only here, so why should we have
>>>> enable parameter in this op?
>>>
>>> I'm not sure if I understand the question, but I will try to answer.
>>>
>>> 'enable' is passed through is31fl32xx_software_shutdown to
>>> cdef->sw_shutdown_func, so it can be either true or false at that
>>> point. The purpose of sw_shutdown_func is to add any special behavior
>>> when enabling/disabling software-shutdown mode, which is needed for
>>> the 3216 because its SSD bit is in a different position and with
>>> opposite polarity.
>>>
>>> Is it that 'enable' in that line of code makes it look like it's being
>>> called with an hardcoded value rather than a variable? If so, perhaps a
>>> different parameter name would make it more obvious? Or a kerneldoc
>>> comment for the function to describe the parameter?
>>>
>>> Or have I totally missed the point of the question?
>>
>> Actually I should have placed this question next to
>> the call to s31fl32xx_software_shutdown() in is31fl32xx_init_regs(),
>> which is passed "false" in the second argument, and there is no
>> other call to s31fl32xx_software_shutdown() in the driver.
>>
>> Having the argument makes people wondering that there is some
>> use case in the driver, where "true" is passed, but it seems not
>> to be the case.
>
> Thanks for the clarification.
>
> Yes, there is currently no explicit call to enable software-shutdown
> mode. Since the reset state of these devices is to have software-shutdown
> enabled, the only explicit operation on it that's currently needed is
> to disable it.
>
> Reasons I can think to have the 'enable' parameter anyways include:
> - It seems logical to me that an API to manipulate the software-shutdown
>    state should allow both enabling and disabling.
> - Having a parameter for that seemed the most obvious way to go, and it
>    was trivial to implement.
> - Alternatively having separate "enable" and "disable" functions would
>    duplicate most of the logic, vs the parameter being just a single
>    conditional. And that would also imply two function pointers in the
>    chipdefs, which I'd prefer to minimize.
> - If anyone wanted to implement suspend/resume in the future, they would
>    most likely do it by enabling/disabling software-shutdown. Supporting
>    both enable/disable from the start should make that trivial to do.

Suspend/resume callbacks are already implemented in led-class.c and
related ops indirectly call brightness_set. If you want to support
suspend/resume you have to set LED_CORE_SUSPENDRESUME flag.
Setting brightness to 0 is equivalent to turning the LED controller
in a shutdown mode, provided that all sub-LEDs are off.

> - I thought the code read better with a bool parameter, vs a longer
>    function name.
>
> So nothing really critical, but mostly just my aesthetic and preference.
>
> Also, I expect that is31fl32xx_software_shutdown() would be inlined, so
> the conditional check in there is optimized out anyways, and there is no
> performance penalty. Looking at the disassembly in my ARMv7a build, both
> of those have indeed happened there.

Not only the size of a binary and the performance, but also code
readability matters. The function is called only with false argument,
so I expect that some people may submit patches optimizing this.

Let's avoid the confusion.

>>>>> +	return 0;
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
>>>>> +{
>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>> +	int ret;
>>>>> +
>>>>> +	ret = is31fl32xx_reset_regs(priv);
>>>>> +	if (ret)
>>>>> +		return ret;
>>>>> +
>>>>> +	/*
>>>>> +	 * Set enable bit for all channels.
>>>>> +	 * We will control state with PWM registers alone.
>>>>> +	 */
>>>>> +	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
>>>>> +		u8 value =
>>>>> +		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
>>>>> +		u8 num_regs = cdef->channels /
>>>>> +				cdef->enable_bits_per_led_control_register;
>>>>> +		int i;
>>>>> +
>>>>> +		for (i = 0; i < num_regs; i++) {
>>>>> +			ret = is31fl32xx_write(priv,
>>>>> +					       cdef->led_control_register_base+i,
>>>>> +					       value);
>>>>> +			if (ret)
>>>>> +				return ret;
>>>>> +		}
>>>>> +	}
>>>>> +
>>>>> +	ret = is31fl32xx_software_shutdown(priv, false);
>>>>> +	if (ret)
>>>>> +		return ret;
>>>>> +
>>>>> +	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
>>>>> +		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
>>>>> +		if (ret)
>>>>> +			return ret;
>>>>> +	}
>>>>> +
>>>>> +	return 0;
>>>>> +}
>>>>> +
>>>>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
>>>>> +{
>>>>> +	return sizeof(struct is31fl32xx_priv) +
>>>>> +		      (sizeof(struct is31fl32xx_led_data) * num_leds);
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
>>>>> +				     const struct device_node *child,
>>>>> +				     struct is31fl32xx_led_data *led_data)
>>>>> +{
>>>>> +	struct led_classdev *cdev = &led_data->cdev;
>>>>> +	int ret = 0;
>>>>> +	u32 reg;
>>>>> +
>>>>> +	if (of_property_read_string(child, "label", &cdev->name))
>>>>> +		cdev->name = child->name;
>>>>> +
>>>>> +	ret = of_property_read_u32(child, "reg", &reg);
>>>>> +	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
>>>>> +		dev_err(dev,
>>>>> +			"Child node %s does not have a valid reg property\n",
>>>>> +			child->full_name);
>>>>> +		return -EINVAL;
>>>>> +	}
>>>>> +	led_data->channel = reg;
>>>>> +
>>>>> +	of_property_read_string(child, "linux,default-trigger",
>>>>> +				&cdev->default_trigger);
>>>>> +
>>>>> +	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
>>>>> +
>>>>> +	return 0;
>>>>> +}
>>>>> +
>>>>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
>>>>> +					struct is31fl32xx_priv *priv,
>>>>> +					u8 channel)
>>>>> +{
>>>>> +	size_t i;
>>>>> +
>>>>> +	for (i = 0; i < priv->num_leds; i++) {
>>>>> +		if (priv->leds[i].channel == channel)
>>>>> +			return &priv->leds[i];
>>>>> +	}
>>>>> +
>>>>> +	return NULL;
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_parse_dt(struct device *dev,
>>>>> +			       struct is31fl32xx_priv *priv)
>>>>> +{
>>>>> +	struct device_node *child;
>>>>> +	int ret = 0;
>>>>> +
>>>>> +	for_each_child_of_node(dev->of_node, child) {
>>>>> +		struct is31fl32xx_led_data *led_data =
>>>>> +			&priv->leds[priv->num_leds];
>>>>> +		const struct is31fl32xx_led_data *other_led_data;
>>>>> +
>>>>> +		led_data->priv = priv;
>>>>> +
>>>>> +		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
>>>>> +		if (ret)
>>>>> +			goto err;
>>>>> +
>>>>> +		/* Detect if channel is already in use by another child */
>>>>> +		other_led_data = is31fl32xx_find_led_data(priv,
>>>>> +							  led_data->channel);
>>>>> +		if (other_led_data) {
>>>>> +			dev_err(dev,
>>>>> +				"%s and %s both attempting to use channel %d\n",
>>>>> +				led_data->cdev.name,
>>>>> +				other_led_data->cdev.name,
>>>>> +				led_data->channel);
>>>>> +			goto err;
>>>>> +		}
>>>>> +
>>>>> +		ret = devm_led_classdev_register(dev, &led_data->cdev);
>>>>> +		if (ret) {
>>>>> +			dev_err(dev, "failed to register PWM led for %s: %d\n",
>>>>> +				led_data->cdev.name, ret);
>>>>> +			goto err;
>>>>> +		}
>>>>> +
>>>>> +		priv->num_leds++;
>>>>> +	}
>>>>> +
>>>>> +	return 0;
>>>>> +
>>>>> +err:
>>>>> +	of_node_put(child);
>>>>> +	return ret;
>>>>> +}
>>>>> +
>>>>> +static const struct of_device_id of_is31fl31xx_match[] = {
>>>>> +	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
>>>>> +	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
>>>>> +	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
>>>>> +	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
>>>>> +	{},
>>>>> +};
>>>>> +
>>>>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
>>>>> +
>>>>> +static int is31fl32xx_probe(struct i2c_client *client,
>>>>> +			    const struct i2c_device_id *id)
>>>>> +{
>>>>> +	const struct is31fl32xx_chipdef *cdef;
>>>>> +	const struct of_device_id *of_dev_id;
>>>>> +	struct device *dev = &client->dev;
>>>>> +	struct is31fl32xx_priv *priv;
>>>>> +	int count;
>>>>> +	int ret = 0;
>>>>> +
>>>>> +	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
>>>>> +	if (!of_dev_id)
>>>>> +		return -EINVAL;
>>>>> +
>>>>> +	cdef = of_dev_id->data;
>>>>> +
>>>>> +	count = of_get_child_count(dev->of_node);
>>>>> +	if (!count)
>>>>> +		return -EINVAL;
>>>>> +
>>>>> +	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
>>>>> +			    GFP_KERNEL);
>>>>> +	if (!priv)
>>>>> +		return -ENOMEM;
>>>>> +
>>>>> +	priv->client = client;
>>>>> +	priv->cdef = cdef;
>>>>> +	i2c_set_clientdata(client, priv);
>>>>> +
>>>>> +	ret = is31fl32xx_init_regs(priv);
>>>>> +	if (ret)
>>>>> +		return ret;
>>>>> +
>>>>> +	ret = is31fl32xx_parse_dt(dev, priv);
>>>>> +	if (ret)
>>>>> +		return ret;
>>>>> +
>>>>> +	return 0;
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_remove(struct i2c_client *client)
>>>>> +{
>>>>> +	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
>>>>> +
>>>>> +	return is31fl32xx_reset_regs(priv);
>>>>> +}
>>>>> +
>>>>> +/*
>>>>> + * i2c-core requires that id_table be non-NULL, even though
>>>>> + * it is not used for DeviceTree based instantiation.
>>>>> + */
>>>>> +static const struct i2c_device_id is31fl31xx_id[] = {
>>>>> +	{},
>>>>> +};
>>>>> +
>>>>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
>>>>> +
>>>>> +static struct i2c_driver is31fl32xx_driver = {
>>>>> +	.driver = {
>>>>> +		.name	= "is31fl32xx",
>>>>> +		.of_match_table = of_is31fl31xx_match,
>>>>> +	},
>>>>> +	.probe		= is31fl32xx_probe,
>>>>> +	.remove		= is31fl32xx_remove,
>>>>> +	.id_table	= is31fl31xx_id,
>>>>> +};
>>>>> +
>>>>> +module_i2c_driver(is31fl32xx_driver);
>>>>> +
>>>>> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
>>>>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
>>>>> +MODULE_LICENSE("GPL v2");
>>>>>
>
>
>


-- 
Best regards,
Jacek Anaszewski

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[David Rivshin (Allworx)]

On Fri, 04 Mar 2016 17:01:52 +0100
Jacek Anaszewski <j.anaszewski@samsung.com> wrote:

> On 03/04/2016 04:05 PM, David Rivshin (Allworx) wrote:
> > On Fri, 04 Mar 2016 08:54:02 +0100
> > Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
> >  
> >> On 03/04/2016 01:45 AM, David Rivshin (Allworx) wrote:  
> >>> On Thu, 03 Mar 2016 15:51:32 +0100
> >>> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
> >>>  
> >>>> Hi David,
> >>>>
> >>>> Thanks for the update. Two remarks in the code.
> >>>>
> >>>> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:  
> >>>>> From: David Rivshin <drivshin@allworx.com>
> >>>>>
> >>>>> The IS31FL32xx family of LED controllers are I2C devices with multiple
> >>>>> constant-current channels, each with independent 256-level PWM control.
> >>>>>
> >>>>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >>>>>
> >>>>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
> >>>>> (TI am335x) platform.
> >>>>>
> >>>>> The programming paradigm of these devices is similar in the following
> >>>>> ways:
> >>>>>     - All registers are 8 bit
> >>>>>     - All LED control registers are write-only
> >>>>>     - Each LED channel has a PWM register (0-255)
> >>>>>     - PWM register writes are shadowed until an Update register is poked
> >>>>>     - All have a concept of Software Shutdown, which disables output
> >>>>>
> >>>>> However, there are some differences in devices:
> >>>>>     - 3236/3235 have a separate Control register for each LED,
> >>>>>       (3218/3216 pack the enable bits into fewer registers)
> >>>>>     - 3236/3235 have a per-channel current divisor setting
> >>>>>     - 3236/3235 have a Global Control register that can turn off all LEDs
> >>>>>     - 3216 is unique in a number of ways
> >>>>>        - OUT9-OUT16 can be configured as GPIOs instead of LED controls
> >>>>>        - LEDs can be programmed with an 8-frame animation, with
> >>>>>          programmable delay between frames
> >>>>>        - LEDs can be modulated by an input audio signal
> >>>>>        - Max output current can be adjusted from 1/4 to 2x globally
> >>>>>        - Has a Configuration register instead of a Shutdown register
> >>>>>
> >>>>> This driver currently only supports the base PWM control function
> >>>>> of these devices. The following features of these devices are not
> >>>>> implemented, although it should be possible to add them in the future:
> >>>>>     - All devices are capable of going into a lower-power "software
> >>>>>       shutdown" mode.
> >>>>>     - The is31fl3236 and is31fl3235 can reduce the max output current
> >>>>>       per-channel with a divisor of 1, 2, 3, or 4.
> >>>>>     - The is31fl3216 can use some LED channels as GPIOs instead.
> >>>>>     - The is31fl3216 can animate LEDs in hardware.
> >>>>>     - The is31fl3216 can modulate LEDs according to an audio input.
> >>>>>     - The is31fl3216 can reduce/increase max output current globally.
> >>>>>
> >>>>> Signed-off-by: David Rivshin <drivshin@allworx.com>
> >>>>> ---
> >>>>>
> >>>>> You may see two instances of this warning:
> >>>>>      "passing argument 1 of 'of_property_read_string' discards 'const'
> >>>>>       qualifier from pointer target type"
> >>>>> That is a result of of_property_read_string() taking a non-const
> >>>>> struct device_node pointer parameter. I have separately submitted a
> >>>>> patch to fix that [1], and a few related functions which had the same
> >>>>> issue. I'm hoping that will get into linux-next before this does, so
> >>>>> that the warnings never show up there.  
> >>>>
> >>>> Please adjust the patch so that it compiles without warnings on
> >>>> current linux-next. Your patch for DT API hasn't been reviewed yet
> >>>> AFICS, and I can imagine that there will be some resistance against.  
> >>>
> >>> Since the DT API patch was just accepted by Rob [1], would it be OK
> >>> to wait for the results of Stefan's testing (and any other reviews)
> >>> before making a decision on this? From Stefan's note, it won't be
> >>> until this weekend that he will have a chance to test, and I'm
> >>> guessing the DT API patch will make its way through Rob's tree to
> >>> linux-next by then.  
> >>
> >> OK.
> >>  
> >>> FYI, the warning workaround would be to make the second parameter to
> >>> is31fl32xx_parse_child_dt() non-const.
> >>>
> >>> [1] https://lkml.org/lkml/2016/3/3/924
> >>>  
> >>>>> Changes from RFC:
> >>>>>     - Removed max-brightness DT property.
> >>>>>     - Refer to these devices as "LED controllers" in Kconfig.
> >>>>>     - Removed redundant last sentence from Kconfig entry
> >>>>>     - Removed unnecessary debug code.
> >>>>>     - Do not set led_classdev.brightness to 0 explicitly, as it is
> >>>>>       already initialized to 0 by devm_kzalloc().
> >>>>>     - Used of_property_read_string() instead of of_get_property().
> >>>>>     - Fail immediately on DT parsing error in a child node, rather than
> >>>>>       continuing on with the non-faulty ones.
> >>>>>     - Added additional comments for some things that might be non-obvious.
> >>>>>     - Added constants for the location of the SSD bit in the SHUTDOWN
> >>>>>       register, and the 3216's CONFIG register.
> >>>>>     - Added special sw_shutdown_func for the 3216 device, as that bit
> >>>>>       is in a different register, at a different position, and has reverse
> >>>>>       polarity compared to all the other devices.
> >>>>>     - Refactored is31fl32xx_init_regs() to separate out some logic into
> >>>>>       is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
> >>>>>
> >>>>> [1] https://lkml.org/lkml/2016/3/2/746
> >>>>>
> >>>>>     drivers/leds/Kconfig           |   8 +
> >>>>>     drivers/leds/Makefile          |   1 +
> >>>>>     drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
> >>>>>     3 files changed, 514 insertions(+)
> >>>>>     create mode 100644 drivers/leds/leds-is31fl32xx.c
> >>>>>
> >>>>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> >>>>> index 1034696..9c63ba4 100644
> >>>>> --- a/drivers/leds/Kconfig
> >>>>> +++ b/drivers/leds/Kconfig
> >>>>> @@ -580,6 +580,14 @@ config LEDS_SN3218
> >>>>>     	  This driver can also be built as a module. If so the module
> >>>>>     	  will be called leds-sn3218.
> >>>>>
> >>>>> +config LEDS_IS31FL32XX
> >>>>> +	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
> >>>>> +	depends on LEDS_CLASS && I2C && OF
> >>>>> +	help
> >>>>> +	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
> >>>>> +	  They are I2C devices with multiple constant-current channels, each
> >>>>> +	  with independent 256-level PWM control.
> >>>>> +
> >>>>>     comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
> >>>>>
> >>>>>     config LEDS_BLINKM
> >>>>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
> >>>>> index 89c9b6f..3fdf313 100644
> >>>>> --- a/drivers/leds/Makefile
> >>>>> +++ b/drivers/leds/Makefile
> >>>>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
> >>>>>     obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
> >>>>>     obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
> >>>>>     obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
> >>>>> +obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
> >>>>>
> >>>>>     # LED SPI Drivers
> >>>>>     obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
> >>>>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
> >>>>> new file mode 100644
> >>>>> index 0000000..49818f0
> >>>>> --- /dev/null
> >>>>> +++ b/drivers/leds/leds-is31fl32xx.c
> >>>>> @@ -0,0 +1,505 @@
> >>>>> +/*
> >>>>> + * linux/drivers/leds-is31fl32xx.c
> >>>>> + *
> >>>>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
> >>>>> + *
> >>>>> + * Copyright 2015 Allworx Corp.
> >>>>> + *
> >>>>> + *
> >>>>> + * 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.
> >>>>> + *
> >>>>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >>>>> + */
> >>>>> +
> >>>>> +#include <linux/err.h>
> >>>>> +#include <linux/i2c.h>
> >>>>> +#include <linux/kernel.h>
> >>>>> +#include <linux/leds.h>
> >>>>> +#include <linux/module.h>
> >>>>> +#include <linux/of_platform.h>
> >>>>> +
> >>>>> +/* Used to indicate a device has no such register */
> >>>>> +#define IS31FL32XX_REG_NONE 0xFF
> >>>>> +
> >>>>> +/* Software Shutdown bit in Shutdown Register */
> >>>>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
> >>>>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
> >>>>> +
> >>>>> +/* IS31FL3216 has a number of unique registers */
> >>>>> +#define IS31FL3216_CONFIG_REG 0x00
> >>>>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
> >>>>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
> >>>>> +
> >>>>> +/* Software Shutdown bit in 3216 Config Register */
> >>>>> +#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
> >>>>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
> >>>>> +
> >>>>> +struct is31fl32xx_priv;
> >>>>> +struct is31fl32xx_led_data {
> >>>>> +	struct led_classdev cdev;
> >>>>> +	u8 channel; /* 1-based, max priv->cdef->channels */
> >>>>> +	struct is31fl32xx_priv *priv;
> >>>>> +};
> >>>>> +
> >>>>> +struct is31fl32xx_priv {
> >>>>> +	const struct is31fl32xx_chipdef *cdef;
> >>>>> +	struct i2c_client *client;
> >>>>> +	unsigned int num_leds;
> >>>>> +	struct is31fl32xx_led_data leds[0];
> >>>>> +};
> >>>>> +
> >>>>> +/**
> >>>>> + * struct is31fl32xx_chipdef - chip-specific attributes
> >>>>> + * @channels            : Number of LED channels
> >>>>> + * @shutdown_reg        : address of Shutdown register (optional)
> >>>>> + * @pwm_update_reg      : address of PWM Update register
> >>>>> + * @global_control_reg  : address of Global Control register (optional)
> >>>>> + * @reset_reg           : address of Reset register (optional)
> >>>>> + * @pwm_register_base   : address of first PWM register
> >>>>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
> >>>>> + * @led_control_register_base : address of first LED control register (optional)
> >>>>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
> >>>>> + * @reset_func:         : pointer to reset function
> >>>>> + *
> >>>>> + * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
> >>>>> + * indicates that this chip has no such register.
> >>>>> + *
> >>>>> + * If non-NULL, @reset_func will be called during probing to set all
> >>>>> + * necessary registers to a known initialization state. This is needed
> >>>>> + * for chips that do not have a @reset_reg.
> >>>>> + *
> >>>>> + * @enable_bits_per_led_control_register must be >=1 if
> >>>>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
> >>>>> + */
> >>>>> +struct is31fl32xx_chipdef {
> >>>>> +	u8	channels;
> >>>>> +	u8	shutdown_reg;
> >>>>> +	u8	pwm_update_reg;
> >>>>> +	u8	global_control_reg;
> >>>>> +	u8	reset_reg;
> >>>>> +	u8	pwm_register_base;
> >>>>> +	bool	pwm_registers_reversed;
> >>>>> +	u8	led_control_register_base;
> >>>>> +	u8	enable_bits_per_led_control_register;
> >>>>> +	int (*reset_func)(struct is31fl32xx_priv *priv);
> >>>>> +	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
> >>>>> +};
> >>>>> +
> >>>>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
> >>>>> +	.channels				= 36,
> >>>>> +	.shutdown_reg				= 0x00,
> >>>>> +	.pwm_update_reg				= 0x25,
> >>>>> +	.global_control_reg			= 0x4a,
> >>>>> +	.reset_reg				= 0x4f,
> >>>>> +	.pwm_register_base			= 0x01,
> >>>>> +	.led_control_register_base		= 0x26,
> >>>>> +	.enable_bits_per_led_control_register	= 1,
> >>>>> +};
> >>>>> +
> >>>>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
> >>>>> +	.channels				= 28,
> >>>>> +	.shutdown_reg				= 0x00,
> >>>>> +	.pwm_update_reg				= 0x25,
> >>>>> +	.global_control_reg			= 0x4a,
> >>>>> +	.reset_reg				= 0x4f,
> >>>>> +	.pwm_register_base			= 0x05,
> >>>>> +	.led_control_register_base		= 0x2a,
> >>>>> +	.enable_bits_per_led_control_register	= 1,
> >>>>> +};
> >>>>> +
> >>>>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
> >>>>> +	.channels				= 18,
> >>>>> +	.shutdown_reg				= 0x00,
> >>>>> +	.pwm_update_reg				= 0x16,
> >>>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
> >>>>> +	.reset_reg				= 0x17,
> >>>>> +	.pwm_register_base			= 0x01,
> >>>>> +	.led_control_register_base		= 0x13,
> >>>>> +	.enable_bits_per_led_control_register	= 6,
> >>>>> +};
> >>>>> +
> >>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
> >>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> >>>>> +					bool enable);
> >>>>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
> >>>>> +	.channels				= 16,
> >>>>> +	.shutdown_reg				= IS31FL32XX_REG_NONE,
> >>>>> +	.pwm_update_reg				= 0xB0,
> >>>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
> >>>>> +	.reset_reg				= IS31FL32XX_REG_NONE,
> >>>>> +	.pwm_register_base			= 0x10,
> >>>>> +	.pwm_registers_reversed			= true,
> >>>>> +	.led_control_register_base		= 0x01,
> >>>>> +	.enable_bits_per_led_control_register	= 8,
> >>>>> +	.reset_func				= is31fl3216_reset,
> >>>>> +	.sw_shutdown_func			= is31fl3216_software_shutdown,
> >>>>> +};
> >>>>> +
> >>>>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
> >>>>> +{
> >>>>> +	int ret;
> >>>>> +
> >>>>> +	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
> >>>>> +
> >>>>> +	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
> >>>>> +	if (ret) {
> >>>>> +		dev_err(&priv->client->dev,
> >>>>> +			"register write to 0x%02X failed (error %d)",
> >>>>> +			reg, ret);
> >>>>> +	}
> >>>>> +	return ret;
> >>>>> +}
> >>>>> +
> >>>>> +/*
> >>>>> + * Custom reset function for IS31FL3216 because it does not have a RESET
> >>>>> + * register the way that the other IS31FL32xx chips do. We don't bother
> >>>>> + * writing the GPIO and animation registers, because the registers we
> >>>>> + * do write ensure those will have no effect.
> >>>>> + */
> >>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
> >>>>> +{
> >>>>> +	unsigned int i;
> >>>>> +	int ret;
> >>>>> +
> >>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
> >>>>> +			       IS31FL3216_CONFIG_SSD_ENABLE);
> >>>>> +	if (ret)
> >>>>> +		return ret;
> >>>>> +	for (i = 0; i < priv->cdef->channels; i++) {
> >>>>> +		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
> >>>>> +				       0x00);
> >>>>> +		if (ret)
> >>>>> +			return ret;
> >>>>> +	}
> >>>>> +	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
> >>>>> +	if (ret)
> >>>>> +		return ret;
> >>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
> >>>>> +	if (ret)
> >>>>> +		return ret;
> >>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
> >>>>> +	if (ret)
> >>>>> +		return ret;
> >>>>> +
> >>>>> +	return 0;
> >>>>> +}
> >>>>> +
> >>>>> +/*
> >>>>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
> >>>>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
> >>>>> + * We don't bother doing a read/modify/write on the CONFIG register because
> >>>>> + * we only ever use a value of '0' for the other fields in that register.
> >>>>> + */
> >>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> >>>>> +					bool enable)
> >>>>> +{
> >>>>> +	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
> >>>>> +			    IS31FL3216_CONFIG_SSD_DISABLE;
> >>>>> +
> >>>>> +	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
> >>>>> +}
> >>>>> +
> >>>>> +/*
> >>>>> + * NOTE: A mutex is not needed in this function because:
> >>>>> + * - All referenced data is read-only after probe()
> >>>>> + * - The I2C core has a mutex on to protect the bus
> >>>>> + * - There are no read/modify/write operations
> >>>>> + * - Intervening operations between the write of the PWM register
> >>>>> + *   and the Update register are harmless.
> >>>>> + *
> >>>>> + * Example:
> >>>>> + *	PWM_REG_1 write 16
> >>>>> + *	UPDATE_REG write 0
> >>>>> + *	PWM_REG_2 write 128
> >>>>> + *	UPDATE_REG write 0
> >>>>> + *   vs:
> >>>>> + *	PWM_REG_1 write 16
> >>>>> + *	PWM_REG_2 write 128
> >>>>> + *	UPDATE_REG write 0
> >>>>> + *	UPDATE_REG write 0
> >>>>> + * are equivalent. Poking the Update register merely applies all PWM
> >>>>> + * register writes up to that point.
> >>>>> + */
> >>>>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
> >>>>> +				     enum led_brightness brightness)
> >>>>> +{
> >>>>> +	const struct is31fl32xx_led_data *led_data =
> >>>>> +		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
> >>>>> +	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
> >>>>> +	u8 pwm_register_offset;
> >>>>> +	int ret;
> >>>>> +
> >>>>> +	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
> >>>>> +
> >>>>> +	/* NOTE: led_data->channel is 1-based */
> >>>>> +	if (cdef->pwm_registers_reversed)
> >>>>> +		pwm_register_offset = cdef->channels - led_data->channel;
> >>>>> +	else
> >>>>> +		pwm_register_offset = led_data->channel - 1;
> >>>>> +
> >>>>> +	ret = is31fl32xx_write(led_data->priv,
> >>>>> +			       cdef->pwm_register_base + pwm_register_offset,
> >>>>> +			       brightness);
> >>>>> +	if (ret)
> >>>>> +		return ret;
> >>>>> +
> >>>>> +	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
> >>>>> +}
> >>>>> +
> >>>>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
> >>>>> +{
> >>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>>>> +	int ret;
> >>>>> +
> >>>>> +	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
> >>>>> +		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
> >>>>> +		if (ret)
> >>>>> +			return ret;
> >>>>> +	}
> >>>>> +
> >>>>> +	if (cdef->reset_func)
> >>>>> +		return cdef->reset_func(priv);
> >>>>> +
> >>>>> +	return 0;
> >>>>> +}
> >>>>> +
> >>>>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
> >>>>> +					bool enable)
> >>>>> +{
> >>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>>>> +	int ret;
> >>>>> +
> >>>>> +	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
> >>>>> +		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
> >>>>> +				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
> >>>>> +		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
> >>>>> +		if (ret)
> >>>>> +			return ret;
> >>>>> +	}
> >>>>> +
> >>>>> +	if (cdef->sw_shutdown_func)
> >>>>> +		return cdef->sw_shutdown_func(priv, enable);  
> >>>>
> >>>> You seem to call sw_shutdown_func only here, so why should we have
> >>>> enable parameter in this op?  
> >>>
> >>> I'm not sure if I understand the question, but I will try to answer.
> >>>
> >>> 'enable' is passed through is31fl32xx_software_shutdown to
> >>> cdef->sw_shutdown_func, so it can be either true or false at that
> >>> point. The purpose of sw_shutdown_func is to add any special behavior
> >>> when enabling/disabling software-shutdown mode, which is needed for
> >>> the 3216 because its SSD bit is in a different position and with
> >>> opposite polarity.
> >>>
> >>> Is it that 'enable' in that line of code makes it look like it's being
> >>> called with an hardcoded value rather than a variable? If so, perhaps a
> >>> different parameter name would make it more obvious? Or a kerneldoc
> >>> comment for the function to describe the parameter?
> >>>
> >>> Or have I totally missed the point of the question?  
> >>
> >> Actually I should have placed this question next to
> >> the call to s31fl32xx_software_shutdown() in is31fl32xx_init_regs(),
> >> which is passed "false" in the second argument, and there is no
> >> other call to s31fl32xx_software_shutdown() in the driver.
> >>
> >> Having the argument makes people wondering that there is some
> >> use case in the driver, where "true" is passed, but it seems not
> >> to be the case.  
> >
> > Thanks for the clarification.
> >
> > Yes, there is currently no explicit call to enable software-shutdown
> > mode. Since the reset state of these devices is to have software-shutdown
> > enabled, the only explicit operation on it that's currently needed is
> > to disable it.
> >
> > Reasons I can think to have the 'enable' parameter anyways include:
> > - It seems logical to me that an API to manipulate the software-shutdown
> >    state should allow both enabling and disabling.
> > - Having a parameter for that seemed the most obvious way to go, and it
> >    was trivial to implement.
> > - Alternatively having separate "enable" and "disable" functions would
> >    duplicate most of the logic, vs the parameter being just a single
> >    conditional. And that would also imply two function pointers in the
> >    chipdefs, which I'd prefer to minimize.
> > - If anyone wanted to implement suspend/resume in the future, they would
> >    most likely do it by enabling/disabling software-shutdown. Supporting
> >    both enable/disable from the start should make that trivial to do.  
> 
> Suspend/resume callbacks are already implemented in led-class.c and
> related ops indirectly call brightness_set. If you want to support
> suspend/resume you have to set LED_CORE_SUSPENDRESUME flag.

If I understand correctly, all LED_CORE_SUSPENDRESUME will do is cause
the led core to set the brightness to 0 on suspend (and reverse that 
on resume). I see some drivers use this flag and other do not.
This brings up the question in my mind: how would a driver decide 
whether it is  appropriate for an LED to go dark on suspend? Is that 
just the drivers that know the logical purpose of the LEDs they control?

> Setting brightness to 0 is equivalent to turning the LED controller
> in a shutdown mode, provided that all sub-LEDs are off.

This is not strictly true for these devices. If someone cared very much 
about power usage when suspended they may want to put the LED controller 
into what it calls "shutdown mode" via the SSD bit. I didn't bother mainly 
because I have no need, and also because I wasn't sure how to even trigger 
a suspend in order to test an implementation.

FYI, I just measured it the effect of software-shutdown even with all LEDs 
already off. The difference in current is about 5mA, measured at the 5V 
supply to a 3216 eval board. Not huge, but someone might care. 

> > - I thought the code read better with a bool parameter, vs a longer
> >    function name.
> >
> > So nothing really critical, but mostly just my aesthetic and preference.
> >
> > Also, I expect that is31fl32xx_software_shutdown() would be inlined, so
> > the conditional check in there is optimized out anyways, and there is no
> > performance penalty. Looking at the disassembly in my ARMv7a build, both
> > of those have indeed happened there.  
> 
> Not only the size of a binary and the performance, but also code
> readability matters. The function is called only with false argument,
> so I expect that some people may submit patches optimizing this.
> 
> Let's avoid the confusion.

I guess we just have a difference of opinion on which way is more
readable, which is OK. Unless the above explanation causes you to 
change your mind, I will remove the 'enable' parameter and add a 
"_disable" suffix to both functions. That will also leave the 
	#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
constant unused in the code, should that also be removed? Note that 
the 3216 specific constant
	#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
would still be used in is31fl3216_reset().

> >>>>> +	return 0;
> >>>>> +}
> >>>>> +
> >>>>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
> >>>>> +{
> >>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>>>> +	int ret;
> >>>>> +
> >>>>> +	ret = is31fl32xx_reset_regs(priv);
> >>>>> +	if (ret)
> >>>>> +		return ret;
> >>>>> +
> >>>>> +	/*
> >>>>> +	 * Set enable bit for all channels.
> >>>>> +	 * We will control state with PWM registers alone.
> >>>>> +	 */
> >>>>> +	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
> >>>>> +		u8 value =
> >>>>> +		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
> >>>>> +		u8 num_regs = cdef->channels /
> >>>>> +				cdef->enable_bits_per_led_control_register;
> >>>>> +		int i;
> >>>>> +
> >>>>> +		for (i = 0; i < num_regs; i++) {
> >>>>> +			ret = is31fl32xx_write(priv,
> >>>>> +					       cdef->led_control_register_base+i,
> >>>>> +					       value);
> >>>>> +			if (ret)
> >>>>> +				return ret;
> >>>>> +		}
> >>>>> +	}
> >>>>> +
> >>>>> +	ret = is31fl32xx_software_shutdown(priv, false);
> >>>>> +	if (ret)
> >>>>> +		return ret;
> >>>>> +
> >>>>> +	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
> >>>>> +		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
> >>>>> +		if (ret)
> >>>>> +			return ret;
> >>>>> +	}
> >>>>> +
> >>>>> +	return 0;
> >>>>> +}
> >>>>> +
> >>>>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
> >>>>> +{
> >>>>> +	return sizeof(struct is31fl32xx_priv) +
> >>>>> +		      (sizeof(struct is31fl32xx_led_data) * num_leds);
> >>>>> +}
> >>>>> +
> >>>>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
> >>>>> +				     const struct device_node *child,
> >>>>> +				     struct is31fl32xx_led_data *led_data)
> >>>>> +{
> >>>>> +	struct led_classdev *cdev = &led_data->cdev;
> >>>>> +	int ret = 0;
> >>>>> +	u32 reg;
> >>>>> +
> >>>>> +	if (of_property_read_string(child, "label", &cdev->name))
> >>>>> +		cdev->name = child->name;
> >>>>> +
> >>>>> +	ret = of_property_read_u32(child, "reg", &reg);
> >>>>> +	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
> >>>>> +		dev_err(dev,
> >>>>> +			"Child node %s does not have a valid reg property\n",
> >>>>> +			child->full_name);
> >>>>> +		return -EINVAL;
> >>>>> +	}
> >>>>> +	led_data->channel = reg;
> >>>>> +
> >>>>> +	of_property_read_string(child, "linux,default-trigger",
> >>>>> +				&cdev->default_trigger);
> >>>>> +
> >>>>> +	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
> >>>>> +
> >>>>> +	return 0;
> >>>>> +}
> >>>>> +
> >>>>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
> >>>>> +					struct is31fl32xx_priv *priv,
> >>>>> +					u8 channel)
> >>>>> +{
> >>>>> +	size_t i;
> >>>>> +
> >>>>> +	for (i = 0; i < priv->num_leds; i++) {
> >>>>> +		if (priv->leds[i].channel == channel)
> >>>>> +			return &priv->leds[i];
> >>>>> +	}
> >>>>> +
> >>>>> +	return NULL;
> >>>>> +}
> >>>>> +
> >>>>> +static int is31fl32xx_parse_dt(struct device *dev,
> >>>>> +			       struct is31fl32xx_priv *priv)
> >>>>> +{
> >>>>> +	struct device_node *child;
> >>>>> +	int ret = 0;
> >>>>> +
> >>>>> +	for_each_child_of_node(dev->of_node, child) {
> >>>>> +		struct is31fl32xx_led_data *led_data =
> >>>>> +			&priv->leds[priv->num_leds];
> >>>>> +		const struct is31fl32xx_led_data *other_led_data;
> >>>>> +
> >>>>> +		led_data->priv = priv;
> >>>>> +
> >>>>> +		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
> >>>>> +		if (ret)
> >>>>> +			goto err;
> >>>>> +
> >>>>> +		/* Detect if channel is already in use by another child */
> >>>>> +		other_led_data = is31fl32xx_find_led_data(priv,
> >>>>> +							  led_data->channel);
> >>>>> +		if (other_led_data) {
> >>>>> +			dev_err(dev,
> >>>>> +				"%s and %s both attempting to use channel %d\n",
> >>>>> +				led_data->cdev.name,
> >>>>> +				other_led_data->cdev.name,
> >>>>> +				led_data->channel);
> >>>>> +			goto err;
> >>>>> +		}
> >>>>> +
> >>>>> +		ret = devm_led_classdev_register(dev, &led_data->cdev);
> >>>>> +		if (ret) {
> >>>>> +			dev_err(dev, "failed to register PWM led for %s: %d\n",
> >>>>> +				led_data->cdev.name, ret);
> >>>>> +			goto err;
> >>>>> +		}
> >>>>> +
> >>>>> +		priv->num_leds++;
> >>>>> +	}
> >>>>> +
> >>>>> +	return 0;
> >>>>> +
> >>>>> +err:
> >>>>> +	of_node_put(child);
> >>>>> +	return ret;
> >>>>> +}
> >>>>> +
> >>>>> +static const struct of_device_id of_is31fl31xx_match[] = {
> >>>>> +	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
> >>>>> +	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
> >>>>> +	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
> >>>>> +	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
> >>>>> +	{},
> >>>>> +};
> >>>>> +
> >>>>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
> >>>>> +
> >>>>> +static int is31fl32xx_probe(struct i2c_client *client,
> >>>>> +			    const struct i2c_device_id *id)
> >>>>> +{
> >>>>> +	const struct is31fl32xx_chipdef *cdef;
> >>>>> +	const struct of_device_id *of_dev_id;
> >>>>> +	struct device *dev = &client->dev;
> >>>>> +	struct is31fl32xx_priv *priv;
> >>>>> +	int count;
> >>>>> +	int ret = 0;
> >>>>> +
> >>>>> +	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
> >>>>> +	if (!of_dev_id)
> >>>>> +		return -EINVAL;
> >>>>> +
> >>>>> +	cdef = of_dev_id->data;
> >>>>> +
> >>>>> +	count = of_get_child_count(dev->of_node);
> >>>>> +	if (!count)
> >>>>> +		return -EINVAL;
> >>>>> +
> >>>>> +	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
> >>>>> +			    GFP_KERNEL);
> >>>>> +	if (!priv)
> >>>>> +		return -ENOMEM;
> >>>>> +
> >>>>> +	priv->client = client;
> >>>>> +	priv->cdef = cdef;
> >>>>> +	i2c_set_clientdata(client, priv);
> >>>>> +
> >>>>> +	ret = is31fl32xx_init_regs(priv);
> >>>>> +	if (ret)
> >>>>> +		return ret;
> >>>>> +
> >>>>> +	ret = is31fl32xx_parse_dt(dev, priv);
> >>>>> +	if (ret)
> >>>>> +		return ret;
> >>>>> +
> >>>>> +	return 0;
> >>>>> +}
> >>>>> +
> >>>>> +static int is31fl32xx_remove(struct i2c_client *client)
> >>>>> +{
> >>>>> +	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
> >>>>> +
> >>>>> +	return is31fl32xx_reset_regs(priv);
> >>>>> +}
> >>>>> +
> >>>>> +/*
> >>>>> + * i2c-core requires that id_table be non-NULL, even though
> >>>>> + * it is not used for DeviceTree based instantiation.
> >>>>> + */
> >>>>> +static const struct i2c_device_id is31fl31xx_id[] = {
> >>>>> +	{},
> >>>>> +};
> >>>>> +
> >>>>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
> >>>>> +
> >>>>> +static struct i2c_driver is31fl32xx_driver = {
> >>>>> +	.driver = {
> >>>>> +		.name	= "is31fl32xx",
> >>>>> +		.of_match_table = of_is31fl31xx_match,
> >>>>> +	},
> >>>>> +	.probe		= is31fl32xx_probe,
> >>>>> +	.remove		= is31fl32xx_remove,
> >>>>> +	.id_table	= is31fl31xx_id,
> >>>>> +};
> >>>>> +
> >>>>> +module_i2c_driver(is31fl32xx_driver);
> >>>>> +
> >>>>> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
> >>>>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
> >>>>> +MODULE_LICENSE("GPL v2");
> >>>>>  

Re: [PATCH 4/4] leds: Replace dedicated SN3218 driver with IS31FL32XX driver

[Stefan Wahren]

Hi David,

> "David Rivshin (Allworx)" <drivshin.allworx@gmail.com> hat am 3. März 2016 um
> 04:01 geschrieben:
>
>
> From: David Rivshin <drivshin@allworx.com>
>
> Si-En Technology was acquired by ISSI in 2011, and it appears that
> the IS31FL3218/IS31FL3216 are just rebranded SN3218/SN3216 devices.
> As the IS31FL32XX driver already handles the *3218 devices, there
> is no longer a need for the dedicated SN3218 driver.
>
> Add the "sn,sn3218" and "sn,sn3216" compatible strings into the
> IS31FL32XX driver and binding documentation, and remove the
> leds-sn3218 driver.
>
> Datasheets:
> IS31FL3218: http://www.issi.com/WW/pdf/31FL3218.pdf
> SN3218: http://www.si-en.com/uploadpdf/s2011517171720.pdf
>
> IS31FL3216: http://www.issi.com/WW/pdf/31FL3216.pdf
> SN3216; http://www.si-en.com/uploadpdf/SN3216201152410148.pdf
>
> Signed-off-by: David Rivshin <drivshin@allworx.com>

i tested this patch successfully with a Raspberry Pi and a PiGlow (SN3218).

Tested-by: Stefan Wahren <stefan.wahren@i2se.com>

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[Jacek Anaszewski]

On 03/04/2016 08:02 PM, David Rivshin (Allworx) wrote:
> On Fri, 04 Mar 2016 17:01:52 +0100
> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>
>> On 03/04/2016 04:05 PM, David Rivshin (Allworx) wrote:
>>> On Fri, 04 Mar 2016 08:54:02 +0100
>>> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>>>
>>>> On 03/04/2016 01:45 AM, David Rivshin (Allworx) wrote:
>>>>> On Thu, 03 Mar 2016 15:51:32 +0100
>>>>> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>>>>>
>>>>>> Hi David,
>>>>>>
>>>>>> Thanks for the update. Two remarks in the code.
>>>>>>
>>>>>> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:
>>>>>>> From: David Rivshin <drivshin@allworx.com>
>>>>>>>
>>>>>>> The IS31FL32xx family of LED controllers are I2C devices with multiple
>>>>>>> constant-current channels, each with independent 256-level PWM control.
>>>>>>>
>>>>>>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>>>>>>
>>>>>>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
>>>>>>> (TI am335x) platform.
>>>>>>>
>>>>>>> The programming paradigm of these devices is similar in the following
>>>>>>> ways:
>>>>>>>      - All registers are 8 bit
>>>>>>>      - All LED control registers are write-only
>>>>>>>      - Each LED channel has a PWM register (0-255)
>>>>>>>      - PWM register writes are shadowed until an Update register is poked
>>>>>>>      - All have a concept of Software Shutdown, which disables output
>>>>>>>
>>>>>>> However, there are some differences in devices:
>>>>>>>      - 3236/3235 have a separate Control register for each LED,
>>>>>>>        (3218/3216 pack the enable bits into fewer registers)
>>>>>>>      - 3236/3235 have a per-channel current divisor setting
>>>>>>>      - 3236/3235 have a Global Control register that can turn off all LEDs
>>>>>>>      - 3216 is unique in a number of ways
>>>>>>>         - OUT9-OUT16 can be configured as GPIOs instead of LED controls
>>>>>>>         - LEDs can be programmed with an 8-frame animation, with
>>>>>>>           programmable delay between frames
>>>>>>>         - LEDs can be modulated by an input audio signal
>>>>>>>         - Max output current can be adjusted from 1/4 to 2x globally
>>>>>>>         - Has a Configuration register instead of a Shutdown register
>>>>>>>
>>>>>>> This driver currently only supports the base PWM control function
>>>>>>> of these devices. The following features of these devices are not
>>>>>>> implemented, although it should be possible to add them in the future:
>>>>>>>      - All devices are capable of going into a lower-power "software
>>>>>>>        shutdown" mode.
>>>>>>>      - The is31fl3236 and is31fl3235 can reduce the max output current
>>>>>>>        per-channel with a divisor of 1, 2, 3, or 4.
>>>>>>>      - The is31fl3216 can use some LED channels as GPIOs instead.
>>>>>>>      - The is31fl3216 can animate LEDs in hardware.
>>>>>>>      - The is31fl3216 can modulate LEDs according to an audio input.
>>>>>>>      - The is31fl3216 can reduce/increase max output current globally.
>>>>>>>
>>>>>>> Signed-off-by: David Rivshin <drivshin@allworx.com>
>>>>>>> ---
>>>>>>>
>>>>>>> You may see two instances of this warning:
>>>>>>>       "passing argument 1 of 'of_property_read_string' discards 'const'
>>>>>>>        qualifier from pointer target type"
>>>>>>> That is a result of of_property_read_string() taking a non-const
>>>>>>> struct device_node pointer parameter. I have separately submitted a
>>>>>>> patch to fix that [1], and a few related functions which had the same
>>>>>>> issue. I'm hoping that will get into linux-next before this does, so
>>>>>>> that the warnings never show up there.
>>>>>>
>>>>>> Please adjust the patch so that it compiles without warnings on
>>>>>> current linux-next. Your patch for DT API hasn't been reviewed yet
>>>>>> AFICS, and I can imagine that there will be some resistance against.
>>>>>
>>>>> Since the DT API patch was just accepted by Rob [1], would it be OK
>>>>> to wait for the results of Stefan's testing (and any other reviews)
>>>>> before making a decision on this? From Stefan's note, it won't be
>>>>> until this weekend that he will have a chance to test, and I'm
>>>>> guessing the DT API patch will make its way through Rob's tree to
>>>>> linux-next by then.
>>>>
>>>> OK.
>>>>
>>>>> FYI, the warning workaround would be to make the second parameter to
>>>>> is31fl32xx_parse_child_dt() non-const.
>>>>>
>>>>> [1] https://lkml.org/lkml/2016/3/3/924
>>>>>
>>>>>>> Changes from RFC:
>>>>>>>      - Removed max-brightness DT property.
>>>>>>>      - Refer to these devices as "LED controllers" in Kconfig.
>>>>>>>      - Removed redundant last sentence from Kconfig entry
>>>>>>>      - Removed unnecessary debug code.
>>>>>>>      - Do not set led_classdev.brightness to 0 explicitly, as it is
>>>>>>>        already initialized to 0 by devm_kzalloc().
>>>>>>>      - Used of_property_read_string() instead of of_get_property().
>>>>>>>      - Fail immediately on DT parsing error in a child node, rather than
>>>>>>>        continuing on with the non-faulty ones.
>>>>>>>      - Added additional comments for some things that might be non-obvious.
>>>>>>>      - Added constants for the location of the SSD bit in the SHUTDOWN
>>>>>>>        register, and the 3216's CONFIG register.
>>>>>>>      - Added special sw_shutdown_func for the 3216 device, as that bit
>>>>>>>        is in a different register, at a different position, and has reverse
>>>>>>>        polarity compared to all the other devices.
>>>>>>>      - Refactored is31fl32xx_init_regs() to separate out some logic into
>>>>>>>        is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
>>>>>>>
>>>>>>> [1] https://lkml.org/lkml/2016/3/2/746
>>>>>>>
>>>>>>>      drivers/leds/Kconfig           |   8 +
>>>>>>>      drivers/leds/Makefile          |   1 +
>>>>>>>      drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
>>>>>>>      3 files changed, 514 insertions(+)
>>>>>>>      create mode 100644 drivers/leds/leds-is31fl32xx.c
>>>>>>>
>>>>>>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
>>>>>>> index 1034696..9c63ba4 100644
>>>>>>> --- a/drivers/leds/Kconfig
>>>>>>> +++ b/drivers/leds/Kconfig
>>>>>>> @@ -580,6 +580,14 @@ config LEDS_SN3218
>>>>>>>      	  This driver can also be built as a module. If so the module
>>>>>>>      	  will be called leds-sn3218.
>>>>>>>
>>>>>>> +config LEDS_IS31FL32XX
>>>>>>> +	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
>>>>>>> +	depends on LEDS_CLASS && I2C && OF
>>>>>>> +	help
>>>>>>> +	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
>>>>>>> +	  They are I2C devices with multiple constant-current channels, each
>>>>>>> +	  with independent 256-level PWM control.
>>>>>>> +
>>>>>>>      comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
>>>>>>>
>>>>>>>      config LEDS_BLINKM
>>>>>>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
>>>>>>> index 89c9b6f..3fdf313 100644
>>>>>>> --- a/drivers/leds/Makefile
>>>>>>> +++ b/drivers/leds/Makefile
>>>>>>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
>>>>>>>      obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
>>>>>>>      obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
>>>>>>>      obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
>>>>>>> +obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
>>>>>>>
>>>>>>>      # LED SPI Drivers
>>>>>>>      obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
>>>>>>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
>>>>>>> new file mode 100644
>>>>>>> index 0000000..49818f0
>>>>>>> --- /dev/null
>>>>>>> +++ b/drivers/leds/leds-is31fl32xx.c
>>>>>>> @@ -0,0 +1,505 @@
>>>>>>> +/*
>>>>>>> + * linux/drivers/leds-is31fl32xx.c
>>>>>>> + *
>>>>>>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
>>>>>>> + *
>>>>>>> + * Copyright 2015 Allworx Corp.
>>>>>>> + *
>>>>>>> + *
>>>>>>> + * 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.
>>>>>>> + *
>>>>>>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>>>>>> + */
>>>>>>> +
>>>>>>> +#include <linux/err.h>
>>>>>>> +#include <linux/i2c.h>
>>>>>>> +#include <linux/kernel.h>
>>>>>>> +#include <linux/leds.h>
>>>>>>> +#include <linux/module.h>
>>>>>>> +#include <linux/of_platform.h>
>>>>>>> +
>>>>>>> +/* Used to indicate a device has no such register */
>>>>>>> +#define IS31FL32XX_REG_NONE 0xFF
>>>>>>> +
>>>>>>> +/* Software Shutdown bit in Shutdown Register */
>>>>>>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
>>>>>>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
>>>>>>> +
>>>>>>> +/* IS31FL3216 has a number of unique registers */
>>>>>>> +#define IS31FL3216_CONFIG_REG 0x00
>>>>>>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
>>>>>>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
>>>>>>> +
>>>>>>> +/* Software Shutdown bit in 3216 Config Register */
>>>>>>> +#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
>>>>>>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
>>>>>>> +
>>>>>>> +struct is31fl32xx_priv;
>>>>>>> +struct is31fl32xx_led_data {
>>>>>>> +	struct led_classdev cdev;
>>>>>>> +	u8 channel; /* 1-based, max priv->cdef->channels */
>>>>>>> +	struct is31fl32xx_priv *priv;
>>>>>>> +};
>>>>>>> +
>>>>>>> +struct is31fl32xx_priv {
>>>>>>> +	const struct is31fl32xx_chipdef *cdef;
>>>>>>> +	struct i2c_client *client;
>>>>>>> +	unsigned int num_leds;
>>>>>>> +	struct is31fl32xx_led_data leds[0];
>>>>>>> +};
>>>>>>> +
>>>>>>> +/**
>>>>>>> + * struct is31fl32xx_chipdef - chip-specific attributes
>>>>>>> + * @channels            : Number of LED channels
>>>>>>> + * @shutdown_reg        : address of Shutdown register (optional)
>>>>>>> + * @pwm_update_reg      : address of PWM Update register
>>>>>>> + * @global_control_reg  : address of Global Control register (optional)
>>>>>>> + * @reset_reg           : address of Reset register (optional)
>>>>>>> + * @pwm_register_base   : address of first PWM register
>>>>>>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
>>>>>>> + * @led_control_register_base : address of first LED control register (optional)
>>>>>>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
>>>>>>> + * @reset_func:         : pointer to reset function
>>>>>>> + *
>>>>>>> + * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
>>>>>>> + * indicates that this chip has no such register.
>>>>>>> + *
>>>>>>> + * If non-NULL, @reset_func will be called during probing to set all
>>>>>>> + * necessary registers to a known initialization state. This is needed
>>>>>>> + * for chips that do not have a @reset_reg.
>>>>>>> + *
>>>>>>> + * @enable_bits_per_led_control_register must be >=1 if
>>>>>>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
>>>>>>> + */
>>>>>>> +struct is31fl32xx_chipdef {
>>>>>>> +	u8	channels;
>>>>>>> +	u8	shutdown_reg;
>>>>>>> +	u8	pwm_update_reg;
>>>>>>> +	u8	global_control_reg;
>>>>>>> +	u8	reset_reg;
>>>>>>> +	u8	pwm_register_base;
>>>>>>> +	bool	pwm_registers_reversed;
>>>>>>> +	u8	led_control_register_base;
>>>>>>> +	u8	enable_bits_per_led_control_register;
>>>>>>> +	int (*reset_func)(struct is31fl32xx_priv *priv);
>>>>>>> +	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
>>>>>>> +};
>>>>>>> +
>>>>>>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
>>>>>>> +	.channels				= 36,
>>>>>>> +	.shutdown_reg				= 0x00,
>>>>>>> +	.pwm_update_reg				= 0x25,
>>>>>>> +	.global_control_reg			= 0x4a,
>>>>>>> +	.reset_reg				= 0x4f,
>>>>>>> +	.pwm_register_base			= 0x01,
>>>>>>> +	.led_control_register_base		= 0x26,
>>>>>>> +	.enable_bits_per_led_control_register	= 1,
>>>>>>> +};
>>>>>>> +
>>>>>>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
>>>>>>> +	.channels				= 28,
>>>>>>> +	.shutdown_reg				= 0x00,
>>>>>>> +	.pwm_update_reg				= 0x25,
>>>>>>> +	.global_control_reg			= 0x4a,
>>>>>>> +	.reset_reg				= 0x4f,
>>>>>>> +	.pwm_register_base			= 0x05,
>>>>>>> +	.led_control_register_base		= 0x2a,
>>>>>>> +	.enable_bits_per_led_control_register	= 1,
>>>>>>> +};
>>>>>>> +
>>>>>>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
>>>>>>> +	.channels				= 18,
>>>>>>> +	.shutdown_reg				= 0x00,
>>>>>>> +	.pwm_update_reg				= 0x16,
>>>>>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
>>>>>>> +	.reset_reg				= 0x17,
>>>>>>> +	.pwm_register_base			= 0x01,
>>>>>>> +	.led_control_register_base		= 0x13,
>>>>>>> +	.enable_bits_per_led_control_register	= 6,
>>>>>>> +};
>>>>>>> +
>>>>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
>>>>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>>>>>> +					bool enable);
>>>>>>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
>>>>>>> +	.channels				= 16,
>>>>>>> +	.shutdown_reg				= IS31FL32XX_REG_NONE,
>>>>>>> +	.pwm_update_reg				= 0xB0,
>>>>>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
>>>>>>> +	.reset_reg				= IS31FL32XX_REG_NONE,
>>>>>>> +	.pwm_register_base			= 0x10,
>>>>>>> +	.pwm_registers_reversed			= true,
>>>>>>> +	.led_control_register_base		= 0x01,
>>>>>>> +	.enable_bits_per_led_control_register	= 8,
>>>>>>> +	.reset_func				= is31fl3216_reset,
>>>>>>> +	.sw_shutdown_func			= is31fl3216_software_shutdown,
>>>>>>> +};
>>>>>>> +
>>>>>>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
>>>>>>> +{
>>>>>>> +	int ret;
>>>>>>> +
>>>>>>> +	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
>>>>>>> +
>>>>>>> +	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
>>>>>>> +	if (ret) {
>>>>>>> +		dev_err(&priv->client->dev,
>>>>>>> +			"register write to 0x%02X failed (error %d)",
>>>>>>> +			reg, ret);
>>>>>>> +	}
>>>>>>> +	return ret;
>>>>>>> +}
>>>>>>> +
>>>>>>> +/*
>>>>>>> + * Custom reset function for IS31FL3216 because it does not have a RESET
>>>>>>> + * register the way that the other IS31FL32xx chips do. We don't bother
>>>>>>> + * writing the GPIO and animation registers, because the registers we
>>>>>>> + * do write ensure those will have no effect.
>>>>>>> + */
>>>>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
>>>>>>> +{
>>>>>>> +	unsigned int i;
>>>>>>> +	int ret;
>>>>>>> +
>>>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
>>>>>>> +			       IS31FL3216_CONFIG_SSD_ENABLE);
>>>>>>> +	if (ret)
>>>>>>> +		return ret;
>>>>>>> +	for (i = 0; i < priv->cdef->channels; i++) {
>>>>>>> +		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
>>>>>>> +				       0x00);
>>>>>>> +		if (ret)
>>>>>>> +			return ret;
>>>>>>> +	}
>>>>>>> +	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
>>>>>>> +	if (ret)
>>>>>>> +		return ret;
>>>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
>>>>>>> +	if (ret)
>>>>>>> +		return ret;
>>>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
>>>>>>> +	if (ret)
>>>>>>> +		return ret;
>>>>>>> +
>>>>>>> +	return 0;
>>>>>>> +}
>>>>>>> +
>>>>>>> +/*
>>>>>>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
>>>>>>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
>>>>>>> + * We don't bother doing a read/modify/write on the CONFIG register because
>>>>>>> + * we only ever use a value of '0' for the other fields in that register.
>>>>>>> + */
>>>>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>>>>>> +					bool enable)
>>>>>>> +{
>>>>>>> +	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
>>>>>>> +			    IS31FL3216_CONFIG_SSD_DISABLE;
>>>>>>> +
>>>>>>> +	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
>>>>>>> +}
>>>>>>> +
>>>>>>> +/*
>>>>>>> + * NOTE: A mutex is not needed in this function because:
>>>>>>> + * - All referenced data is read-only after probe()
>>>>>>> + * - The I2C core has a mutex on to protect the bus
>>>>>>> + * - There are no read/modify/write operations
>>>>>>> + * - Intervening operations between the write of the PWM register
>>>>>>> + *   and the Update register are harmless.
>>>>>>> + *
>>>>>>> + * Example:
>>>>>>> + *	PWM_REG_1 write 16
>>>>>>> + *	UPDATE_REG write 0
>>>>>>> + *	PWM_REG_2 write 128
>>>>>>> + *	UPDATE_REG write 0
>>>>>>> + *   vs:
>>>>>>> + *	PWM_REG_1 write 16
>>>>>>> + *	PWM_REG_2 write 128
>>>>>>> + *	UPDATE_REG write 0
>>>>>>> + *	UPDATE_REG write 0
>>>>>>> + * are equivalent. Poking the Update register merely applies all PWM
>>>>>>> + * register writes up to that point.
>>>>>>> + */
>>>>>>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
>>>>>>> +				     enum led_brightness brightness)
>>>>>>> +{
>>>>>>> +	const struct is31fl32xx_led_data *led_data =
>>>>>>> +		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
>>>>>>> +	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
>>>>>>> +	u8 pwm_register_offset;
>>>>>>> +	int ret;
>>>>>>> +
>>>>>>> +	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
>>>>>>> +
>>>>>>> +	/* NOTE: led_data->channel is 1-based */
>>>>>>> +	if (cdef->pwm_registers_reversed)
>>>>>>> +		pwm_register_offset = cdef->channels - led_data->channel;
>>>>>>> +	else
>>>>>>> +		pwm_register_offset = led_data->channel - 1;
>>>>>>> +
>>>>>>> +	ret = is31fl32xx_write(led_data->priv,
>>>>>>> +			       cdef->pwm_register_base + pwm_register_offset,
>>>>>>> +			       brightness);
>>>>>>> +	if (ret)
>>>>>>> +		return ret;
>>>>>>> +
>>>>>>> +	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
>>>>>>> +}
>>>>>>> +
>>>>>>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
>>>>>>> +{
>>>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>>>> +	int ret;
>>>>>>> +
>>>>>>> +	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
>>>>>>> +		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
>>>>>>> +		if (ret)
>>>>>>> +			return ret;
>>>>>>> +	}
>>>>>>> +
>>>>>>> +	if (cdef->reset_func)
>>>>>>> +		return cdef->reset_func(priv);
>>>>>>> +
>>>>>>> +	return 0;
>>>>>>> +}
>>>>>>> +
>>>>>>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
>>>>>>> +					bool enable)
>>>>>>> +{
>>>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>>>> +	int ret;
>>>>>>> +
>>>>>>> +	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
>>>>>>> +		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
>>>>>>> +				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
>>>>>>> +		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
>>>>>>> +		if (ret)
>>>>>>> +			return ret;
>>>>>>> +	}
>>>>>>> +
>>>>>>> +	if (cdef->sw_shutdown_func)
>>>>>>> +		return cdef->sw_shutdown_func(priv, enable);
>>>>>>
>>>>>> You seem to call sw_shutdown_func only here, so why should we have
>>>>>> enable parameter in this op?
>>>>>
>>>>> I'm not sure if I understand the question, but I will try to answer.
>>>>>
>>>>> 'enable' is passed through is31fl32xx_software_shutdown to
>>>>> cdef->sw_shutdown_func, so it can be either true or false at that
>>>>> point. The purpose of sw_shutdown_func is to add any special behavior
>>>>> when enabling/disabling software-shutdown mode, which is needed for
>>>>> the 3216 because its SSD bit is in a different position and with
>>>>> opposite polarity.
>>>>>
>>>>> Is it that 'enable' in that line of code makes it look like it's being
>>>>> called with an hardcoded value rather than a variable? If so, perhaps a
>>>>> different parameter name would make it more obvious? Or a kerneldoc
>>>>> comment for the function to describe the parameter?
>>>>>
>>>>> Or have I totally missed the point of the question?
>>>>
>>>> Actually I should have placed this question next to
>>>> the call to s31fl32xx_software_shutdown() in is31fl32xx_init_regs(),
>>>> which is passed "false" in the second argument, and there is no
>>>> other call to s31fl32xx_software_shutdown() in the driver.
>>>>
>>>> Having the argument makes people wondering that there is some
>>>> use case in the driver, where "true" is passed, but it seems not
>>>> to be the case.
>>>
>>> Thanks for the clarification.
>>>
>>> Yes, there is currently no explicit call to enable software-shutdown
>>> mode. Since the reset state of these devices is to have software-shutdown
>>> enabled, the only explicit operation on it that's currently needed is
>>> to disable it.
>>>
>>> Reasons I can think to have the 'enable' parameter anyways include:
>>> - It seems logical to me that an API to manipulate the software-shutdown
>>>     state should allow both enabling and disabling.
>>> - Having a parameter for that seemed the most obvious way to go, and it
>>>     was trivial to implement.
>>> - Alternatively having separate "enable" and "disable" functions would
>>>     duplicate most of the logic, vs the parameter being just a single
>>>     conditional. And that would also imply two function pointers in the
>>>     chipdefs, which I'd prefer to minimize.
>>> - If anyone wanted to implement suspend/resume in the future, they would
>>>     most likely do it by enabling/disabling software-shutdown. Supporting
>>>     both enable/disable from the start should make that trivial to do.
>>
>> Suspend/resume callbacks are already implemented in led-class.c and
>> related ops indirectly call brightness_set. If you want to support
>> suspend/resume you have to set LED_CORE_SUSPENDRESUME flag.
>
> If I understand correctly, all LED_CORE_SUSPENDRESUME will do is cause
> the led core to set the brightness to 0 on suspend (and reverse that
> on resume). I see some drivers use this flag and other do not.
> This brings up the question in my mind: how would a driver decide
> whether it is  appropriate for an LED to go dark on suspend? Is that
> just the drivers that know the logical purpose of the LEDs they control?

There is a room for improvement here. Possibly a new LED class sysfs
attribute could be of help in determining that.

>> Setting brightness to 0 is equivalent to turning the LED controller
>> in a shutdown mode, provided that all sub-LEDs are off.
>
> This is not strictly true for these devices. If someone cared very much
> about power usage when suspended they may want to put the LED controller
> into what it calls "shutdown mode" via the SSD bit. I didn't bother mainly
> because I have no need, and also because I wasn't sure how to even trigger
> a suspend in order to test an implementation.

I meant that LED class driver should put the device in a software
shutdown mode after last sub-LED is turned off.

> FYI, I just measured it the effect of software-shutdown even with all LEDs
> already off. The difference in current is about 5mA, measured at the 5V
> supply to a 3216 eval board. Not huge, but someone might care.

This can be vital difference for some use cases. You could count the
number of currently active sub-LEDs and put the controller in a software
shutdown mode in case the value is 0.

>>> - I thought the code read better with a bool parameter, vs a longer
>>>     function name.
>>>
>>> So nothing really critical, but mostly just my aesthetic and preference.
>>>
>>> Also, I expect that is31fl32xx_software_shutdown() would be inlined, so
>>> the conditional check in there is optimized out anyways, and there is no
>>> performance penalty. Looking at the disassembly in my ARMv7a build, both
>>> of those have indeed happened there.
>>
>> Not only the size of a binary and the performance, but also code
>> readability matters. The function is called only with false argument,
>> so I expect that some people may submit patches optimizing this.
>>
>> Let's avoid the confusion.
>
> I guess we just have a difference of opinion on which way is more
> readable, which is OK. Unless the above explanation causes you to
> change your mind, I will remove the 'enable' parameter and add a
> "_disable" suffix to both functions. That will also leave the
> 	#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
> constant unused in the code, should that also be removed? Note that
> the 3216 specific constant
> 	#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
> would still be used in is31fl3216_reset().

Current version of the function would be required for enabling
shutting down the controller from brightness_set op when the
count of active sub-LEDs drops to 0.

>>>>>>> +	return 0;
>>>>>>> +}
>>>>>>> +
>>>>>>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
>>>>>>> +{
>>>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>>>> +	int ret;
>>>>>>> +
>>>>>>> +	ret = is31fl32xx_reset_regs(priv);
>>>>>>> +	if (ret)
>>>>>>> +		return ret;
>>>>>>> +
>>>>>>> +	/*
>>>>>>> +	 * Set enable bit for all channels.
>>>>>>> +	 * We will control state with PWM registers alone.
>>>>>>> +	 */
>>>>>>> +	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
>>>>>>> +		u8 value =
>>>>>>> +		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
>>>>>>> +		u8 num_regs = cdef->channels /
>>>>>>> +				cdef->enable_bits_per_led_control_register;
>>>>>>> +		int i;
>>>>>>> +
>>>>>>> +		for (i = 0; i < num_regs; i++) {
>>>>>>> +			ret = is31fl32xx_write(priv,
>>>>>>> +					       cdef->led_control_register_base+i,
>>>>>>> +					       value);
>>>>>>> +			if (ret)
>>>>>>> +				return ret;
>>>>>>> +		}
>>>>>>> +	}
>>>>>>> +
>>>>>>> +	ret = is31fl32xx_software_shutdown(priv, false);
>>>>>>> +	if (ret)
>>>>>>> +		return ret;
>>>>>>> +
>>>>>>> +	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
>>>>>>> +		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
>>>>>>> +		if (ret)
>>>>>>> +			return ret;
>>>>>>> +	}
>>>>>>> +
>>>>>>> +	return 0;
>>>>>>> +}
>>>>>>> +
>>>>>>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
>>>>>>> +{
>>>>>>> +	return sizeof(struct is31fl32xx_priv) +
>>>>>>> +		      (sizeof(struct is31fl32xx_led_data) * num_leds);
>>>>>>> +}
>>>>>>> +
>>>>>>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
>>>>>>> +				     const struct device_node *child,
>>>>>>> +				     struct is31fl32xx_led_data *led_data)
>>>>>>> +{
>>>>>>> +	struct led_classdev *cdev = &led_data->cdev;
>>>>>>> +	int ret = 0;
>>>>>>> +	u32 reg;
>>>>>>> +
>>>>>>> +	if (of_property_read_string(child, "label", &cdev->name))
>>>>>>> +		cdev->name = child->name;
>>>>>>> +
>>>>>>> +	ret = of_property_read_u32(child, "reg", &reg);
>>>>>>> +	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
>>>>>>> +		dev_err(dev,
>>>>>>> +			"Child node %s does not have a valid reg property\n",
>>>>>>> +			child->full_name);
>>>>>>> +		return -EINVAL;
>>>>>>> +	}
>>>>>>> +	led_data->channel = reg;
>>>>>>> +
>>>>>>> +	of_property_read_string(child, "linux,default-trigger",
>>>>>>> +				&cdev->default_trigger);
>>>>>>> +
>>>>>>> +	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
>>>>>>> +
>>>>>>> +	return 0;
>>>>>>> +}
>>>>>>> +
>>>>>>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
>>>>>>> +					struct is31fl32xx_priv *priv,
>>>>>>> +					u8 channel)
>>>>>>> +{
>>>>>>> +	size_t i;
>>>>>>> +
>>>>>>> +	for (i = 0; i < priv->num_leds; i++) {
>>>>>>> +		if (priv->leds[i].channel == channel)
>>>>>>> +			return &priv->leds[i];
>>>>>>> +	}
>>>>>>> +
>>>>>>> +	return NULL;
>>>>>>> +}
>>>>>>> +
>>>>>>> +static int is31fl32xx_parse_dt(struct device *dev,
>>>>>>> +			       struct is31fl32xx_priv *priv)
>>>>>>> +{
>>>>>>> +	struct device_node *child;
>>>>>>> +	int ret = 0;
>>>>>>> +
>>>>>>> +	for_each_child_of_node(dev->of_node, child) {
>>>>>>> +		struct is31fl32xx_led_data *led_data =
>>>>>>> +			&priv->leds[priv->num_leds];
>>>>>>> +		const struct is31fl32xx_led_data *other_led_data;
>>>>>>> +
>>>>>>> +		led_data->priv = priv;
>>>>>>> +
>>>>>>> +		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
>>>>>>> +		if (ret)
>>>>>>> +			goto err;
>>>>>>> +
>>>>>>> +		/* Detect if channel is already in use by another child */
>>>>>>> +		other_led_data = is31fl32xx_find_led_data(priv,
>>>>>>> +							  led_data->channel);
>>>>>>> +		if (other_led_data) {
>>>>>>> +			dev_err(dev,
>>>>>>> +				"%s and %s both attempting to use channel %d\n",
>>>>>>> +				led_data->cdev.name,
>>>>>>> +				other_led_data->cdev.name,
>>>>>>> +				led_data->channel);
>>>>>>> +			goto err;
>>>>>>> +		}
>>>>>>> +
>>>>>>> +		ret = devm_led_classdev_register(dev, &led_data->cdev);
>>>>>>> +		if (ret) {
>>>>>>> +			dev_err(dev, "failed to register PWM led for %s: %d\n",
>>>>>>> +				led_data->cdev.name, ret);
>>>>>>> +			goto err;
>>>>>>> +		}
>>>>>>> +
>>>>>>> +		priv->num_leds++;
>>>>>>> +	}
>>>>>>> +
>>>>>>> +	return 0;
>>>>>>> +
>>>>>>> +err:
>>>>>>> +	of_node_put(child);
>>>>>>> +	return ret;
>>>>>>> +}
>>>>>>> +
>>>>>>> +static const struct of_device_id of_is31fl31xx_match[] = {
>>>>>>> +	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
>>>>>>> +	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
>>>>>>> +	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
>>>>>>> +	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
>>>>>>> +	{},
>>>>>>> +};
>>>>>>> +
>>>>>>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
>>>>>>> +
>>>>>>> +static int is31fl32xx_probe(struct i2c_client *client,
>>>>>>> +			    const struct i2c_device_id *id)
>>>>>>> +{
>>>>>>> +	const struct is31fl32xx_chipdef *cdef;
>>>>>>> +	const struct of_device_id *of_dev_id;
>>>>>>> +	struct device *dev = &client->dev;
>>>>>>> +	struct is31fl32xx_priv *priv;
>>>>>>> +	int count;
>>>>>>> +	int ret = 0;
>>>>>>> +
>>>>>>> +	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
>>>>>>> +	if (!of_dev_id)
>>>>>>> +		return -EINVAL;
>>>>>>> +
>>>>>>> +	cdef = of_dev_id->data;
>>>>>>> +
>>>>>>> +	count = of_get_child_count(dev->of_node);
>>>>>>> +	if (!count)
>>>>>>> +		return -EINVAL;
>>>>>>> +
>>>>>>> +	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
>>>>>>> +			    GFP_KERNEL);
>>>>>>> +	if (!priv)
>>>>>>> +		return -ENOMEM;
>>>>>>> +
>>>>>>> +	priv->client = client;
>>>>>>> +	priv->cdef = cdef;
>>>>>>> +	i2c_set_clientdata(client, priv);
>>>>>>> +
>>>>>>> +	ret = is31fl32xx_init_regs(priv);
>>>>>>> +	if (ret)
>>>>>>> +		return ret;
>>>>>>> +
>>>>>>> +	ret = is31fl32xx_parse_dt(dev, priv);
>>>>>>> +	if (ret)
>>>>>>> +		return ret;
>>>>>>> +
>>>>>>> +	return 0;
>>>>>>> +}
>>>>>>> +
>>>>>>> +static int is31fl32xx_remove(struct i2c_client *client)
>>>>>>> +{
>>>>>>> +	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
>>>>>>> +
>>>>>>> +	return is31fl32xx_reset_regs(priv);
>>>>>>> +}
>>>>>>> +
>>>>>>> +/*
>>>>>>> + * i2c-core requires that id_table be non-NULL, even though
>>>>>>> + * it is not used for DeviceTree based instantiation.
>>>>>>> + */
>>>>>>> +static const struct i2c_device_id is31fl31xx_id[] = {
>>>>>>> +	{},
>>>>>>> +};
>>>>>>> +
>>>>>>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
>>>>>>> +
>>>>>>> +static struct i2c_driver is31fl32xx_driver = {
>>>>>>> +	.driver = {
>>>>>>> +		.name	= "is31fl32xx",
>>>>>>> +		.of_match_table = of_is31fl31xx_match,
>>>>>>> +	},
>>>>>>> +	.probe		= is31fl32xx_probe,
>>>>>>> +	.remove		= is31fl32xx_remove,
>>>>>>> +	.id_table	= is31fl31xx_id,
>>>>>>> +};
>>>>>>> +
>>>>>>> +module_i2c_driver(is31fl32xx_driver);
>>>>>>> +
>>>>>>> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
>>>>>>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
>>>>>>> +MODULE_LICENSE("GPL v2");
>>>>>>>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-leds" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

-- 
Best Regards,
Jacek Anaszewski

Re: [PATCH 1/4] DT: Add vendor prefix for Integrated Silicon Solutions Inc.

[Rob Herring]

On Wed, Mar 02, 2016 at 10:01:32PM -0500, David Rivshin (Allworx) wrote:
> From: David Rivshin <drivshin@allworx.com>
> 
> ISSI is the stock ticker Integrated Silicon Solutions Inc.
> Company website: http://www.issi.com
> 
> Signed-off-by: David Rivshin <drivshin@allworx.com>
> ---
> 
> Changes from RFC:
>  none
> 
>  Documentation/devicetree/bindings/vendor-prefixes.txt | 1 +
>  1 file changed, 1 insertion(+)

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

Re: [PATCH 2/4] DT: leds: Add binding for the ISSI IS31FL32xx family of LED controllers

[Rob Herring]

On Wed, Mar 02, 2016 at 10:01:33PM -0500, David Rivshin (Allworx) wrote:
> From: David Rivshin <drivshin@allworx.com>
> 
> This adds a binding description for the is31fl3236/35/18/16 I2C LED
> controllers.
> 
> Signed-off-by: David Rivshin <drivshin@allworx.com>
> ---
> 
> Rob,
>  I went with the 1-based 'reg' property here. I inferred that that
> would be your preference based on the previous thread [1]. Let me
> know if that's not the case.
> 
> Changes from RFC:
>  - Removed max-brightness property.
>  - Added #address-cells and #size-cells properties to the example.
> 
> [1] http://www.spinics.net/lists/linux-leds/msg05589.html
> 
>  .../devicetree/bindings/leds/leds-is31fl32xx.txt   | 49 ++++++++++++++++++++++
>  1 file changed, 49 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
> 
> diff --git a/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
> new file mode 100644
> index 0000000..539df2e
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
> @@ -0,0 +1,49 @@
> +Binding for ISSI IS31FL32xx LED Drivers
> +
> +The IS31FL32xx family of LED drivers are I2C devices with multiple
> +constant-current channels, each with independent 256-level PWM control.
> +Each LED is represented as a sub-node of the device.
> +
> +Required properties:
> +- compatible: one of
> +	issi,is31fl3236
> +	issi,is31fl3235
> +	issi,is31fl3218
> +	issi,is31fl3216
> +- reg: I2C slave address
> +- address-cells : must be 1
> +- size-cells : must be 0
> +
> +LED sub-node properties:
> +- reg : LED channel number (1..N)
> +- label :  (optional)
> +  see Documentation/devicetree/bindings/leds/common.txt
> +- linux,default-trigger :  (optional)
> +  see Documentation/devicetree/bindings/leds/common.txt
> +
> +
> +Example:
> +
> +leds: is31fl3236@3c {

Other way around: is31fl3236: leds@3c

Otherwise,

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

Re: [PATCH 4/4] leds: Replace dedicated SN3218 driver with IS31FL32XX driver

[Rob Herring]

On Wed, Mar 02, 2016 at 10:01:35PM -0500, David Rivshin (Allworx) wrote:
> From: David Rivshin <drivshin@allworx.com>
> 
> Si-En Technology was acquired by ISSI in 2011, and it appears that
> the IS31FL3218/IS31FL3216 are just rebranded SN3218/SN3216 devices.
> As the IS31FL32XX driver already handles the *3218 devices, there
> is no longer a need for the dedicated SN3218 driver.
> 
> Add the "sn,sn3218" and "sn,sn3216" compatible strings into the
> IS31FL32XX driver and binding documentation, and remove the
> leds-sn3218 driver.
> 
> Datasheets:
>     IS31FL3218: http://www.issi.com/WW/pdf/31FL3218.pdf
>     SN3218:     http://www.si-en.com/uploadpdf/s2011517171720.pdf
> 
>     IS31FL3216: http://www.issi.com/WW/pdf/31FL3216.pdf
>     SN3216;     http://www.si-en.com/uploadpdf/SN3216201152410148.pdf
> 
> Signed-off-by: David Rivshin <drivshin@allworx.com>
> ---
> 
> Note that the leds-sn3218 binding use a 0-based 'reg' property, while
> the leds-is31fl32xx binding uses a 1-based 'reg' property. This seemed
> to be the preferred binding based on [1]. Since leds-sn3216 has not been
> in a released kernel, there is are no backwards-compatibility concerns.

It would probably make more sense to revert the sn3216 driver/binding 
and add the additions to the previous patches. Doesn't matter to me 
though.

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

> 
> Changes from RFC:
>  new
> 
> [1] http://www.spinics.net/lists/linux-leds/msg05589.html
> 
>  .../devicetree/bindings/leds/leds-is31fl32xx.txt   |   9 +-
>  .../devicetree/bindings/leds/leds-sn3218.txt       |  41 ---
>  drivers/leds/Kconfig                               |  18 +-
>  drivers/leds/Makefile                              |   1 -
>  drivers/leds/leds-is31fl32xx.c                     |   6 +-
>  drivers/leds/leds-sn3218.c                         | 306 ---------------------
>  6 files changed, 14 insertions(+), 367 deletions(-)
>  delete mode 100644 Documentation/devicetree/bindings/leds/leds-sn3218.txt
>  delete mode 100644 drivers/leds/leds-sn3218.c

Re: [PATCH 4/4] leds: Replace dedicated SN3218 driver with IS31FL32XX driver

[David Rivshin (Allworx)]

On Fri, 4 Mar 2016 22:14:27 +0100 (CET)
Stefan Wahren <stefan.wahren@i2se.com> wrote:

> Hi David,
> 
> > "David Rivshin (Allworx)" <drivshin.allworx@gmail.com> hat am 3. März 2016 um
> > 04:01 geschrieben:
> >
> >
> > From: David Rivshin <drivshin@allworx.com>
> >
> > Si-En Technology was acquired by ISSI in 2011, and it appears that
> > the IS31FL3218/IS31FL3216 are just rebranded SN3218/SN3216 devices.
> > As the IS31FL32XX driver already handles the *3218 devices, there
> > is no longer a need for the dedicated SN3218 driver.
> >
> > Add the "sn,sn3218" and "sn,sn3216" compatible strings into the
> > IS31FL32XX driver and binding documentation, and remove the
> > leds-sn3218 driver.
> >
> > Datasheets:
> > IS31FL3218: http://www.issi.com/WW/pdf/31FL3218.pdf
> > SN3218: http://www.si-en.com/uploadpdf/s2011517171720.pdf
> >
> > IS31FL3216: http://www.issi.com/WW/pdf/31FL3216.pdf
> > SN3216; http://www.si-en.com/uploadpdf/SN3216201152410148.pdf
> >
> > Signed-off-by: David Rivshin <drivshin@allworx.com>  
> 
> i tested this patch successfully with a Raspberry Pi and a PiGlow (SN3218).
> 
> Tested-by: Stefan Wahren <stefan.wahren@i2se.com>

Thanks very much for testing!

The PiGlow looks interesting, if I had known about it I might have just
ordered one for testing myself. Looks rather easier to hook up than the 
ISSI 3216 eval board.

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[David Rivshin (Allworx)]

On Fri, 4 Mar 2016 22:39:06 +0100
Jacek Anaszewski <jacek.anaszewski@gmail.com> wrote:

> On 03/04/2016 08:02 PM, David Rivshin (Allworx) wrote:
> > On Fri, 04 Mar 2016 17:01:52 +0100
> > Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
> >  
> >> On 03/04/2016 04:05 PM, David Rivshin (Allworx) wrote:  
> >>> On Fri, 04 Mar 2016 08:54:02 +0100
> >>> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
> >>>  
> >>>> On 03/04/2016 01:45 AM, David Rivshin (Allworx) wrote:  
> >>>>> On Thu, 03 Mar 2016 15:51:32 +0100
> >>>>> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
> >>>>>  
> >>>>>> Hi David,
> >>>>>>
> >>>>>> Thanks for the update. Two remarks in the code.
> >>>>>>
> >>>>>> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:  
> >>>>>>> From: David Rivshin <drivshin@allworx.com>
> >>>>>>>
> >>>>>>> The IS31FL32xx family of LED controllers are I2C devices with multiple
> >>>>>>> constant-current channels, each with independent 256-level PWM control.
> >>>>>>>
> >>>>>>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >>>>>>>
> >>>>>>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
> >>>>>>> (TI am335x) platform.
> >>>>>>>
> >>>>>>> The programming paradigm of these devices is similar in the following
> >>>>>>> ways:
> >>>>>>>      - All registers are 8 bit
> >>>>>>>      - All LED control registers are write-only
> >>>>>>>      - Each LED channel has a PWM register (0-255)
> >>>>>>>      - PWM register writes are shadowed until an Update register is poked
> >>>>>>>      - All have a concept of Software Shutdown, which disables output
> >>>>>>>
> >>>>>>> However, there are some differences in devices:
> >>>>>>>      - 3236/3235 have a separate Control register for each LED,
> >>>>>>>        (3218/3216 pack the enable bits into fewer registers)
> >>>>>>>      - 3236/3235 have a per-channel current divisor setting
> >>>>>>>      - 3236/3235 have a Global Control register that can turn off all LEDs
> >>>>>>>      - 3216 is unique in a number of ways
> >>>>>>>         - OUT9-OUT16 can be configured as GPIOs instead of LED controls
> >>>>>>>         - LEDs can be programmed with an 8-frame animation, with
> >>>>>>>           programmable delay between frames
> >>>>>>>         - LEDs can be modulated by an input audio signal
> >>>>>>>         - Max output current can be adjusted from 1/4 to 2x globally
> >>>>>>>         - Has a Configuration register instead of a Shutdown register
> >>>>>>>
> >>>>>>> This driver currently only supports the base PWM control function
> >>>>>>> of these devices. The following features of these devices are not
> >>>>>>> implemented, although it should be possible to add them in the future:
> >>>>>>>      - All devices are capable of going into a lower-power "software
> >>>>>>>        shutdown" mode.
> >>>>>>>      - The is31fl3236 and is31fl3235 can reduce the max output current
> >>>>>>>        per-channel with a divisor of 1, 2, 3, or 4.
> >>>>>>>      - The is31fl3216 can use some LED channels as GPIOs instead.
> >>>>>>>      - The is31fl3216 can animate LEDs in hardware.
> >>>>>>>      - The is31fl3216 can modulate LEDs according to an audio input.
> >>>>>>>      - The is31fl3216 can reduce/increase max output current globally.
> >>>>>>>
> >>>>>>> Signed-off-by: David Rivshin <drivshin@allworx.com>
> >>>>>>> ---
> >>>>>>>
> >>>>>>> You may see two instances of this warning:
> >>>>>>>       "passing argument 1 of 'of_property_read_string' discards 'const'
> >>>>>>>        qualifier from pointer target type"
> >>>>>>> That is a result of of_property_read_string() taking a non-const
> >>>>>>> struct device_node pointer parameter. I have separately submitted a
> >>>>>>> patch to fix that [1], and a few related functions which had the same
> >>>>>>> issue. I'm hoping that will get into linux-next before this does, so
> >>>>>>> that the warnings never show up there.  
> >>>>>>
> >>>>>> Please adjust the patch so that it compiles without warnings on
> >>>>>> current linux-next. Your patch for DT API hasn't been reviewed yet
> >>>>>> AFICS, and I can imagine that there will be some resistance against.  
> >>>>>
> >>>>> Since the DT API patch was just accepted by Rob [1], would it be OK
> >>>>> to wait for the results of Stefan's testing (and any other reviews)
> >>>>> before making a decision on this? From Stefan's note, it won't be
> >>>>> until this weekend that he will have a chance to test, and I'm
> >>>>> guessing the DT API patch will make its way through Rob's tree to
> >>>>> linux-next by then.  
> >>>>
> >>>> OK.
> >>>>  
> >>>>> FYI, the warning workaround would be to make the second parameter to
> >>>>> is31fl32xx_parse_child_dt() non-const.
> >>>>>
> >>>>> [1] https://lkml.org/lkml/2016/3/3/924
> >>>>>  
> >>>>>>> Changes from RFC:
> >>>>>>>      - Removed max-brightness DT property.
> >>>>>>>      - Refer to these devices as "LED controllers" in Kconfig.
> >>>>>>>      - Removed redundant last sentence from Kconfig entry
> >>>>>>>      - Removed unnecessary debug code.
> >>>>>>>      - Do not set led_classdev.brightness to 0 explicitly, as it is
> >>>>>>>        already initialized to 0 by devm_kzalloc().
> >>>>>>>      - Used of_property_read_string() instead of of_get_property().
> >>>>>>>      - Fail immediately on DT parsing error in a child node, rather than
> >>>>>>>        continuing on with the non-faulty ones.
> >>>>>>>      - Added additional comments for some things that might be non-obvious.
> >>>>>>>      - Added constants for the location of the SSD bit in the SHUTDOWN
> >>>>>>>        register, and the 3216's CONFIG register.
> >>>>>>>      - Added special sw_shutdown_func for the 3216 device, as that bit
> >>>>>>>        is in a different register, at a different position, and has reverse
> >>>>>>>        polarity compared to all the other devices.
> >>>>>>>      - Refactored is31fl32xx_init_regs() to separate out some logic into
> >>>>>>>        is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
> >>>>>>>
> >>>>>>> [1] https://lkml.org/lkml/2016/3/2/746
> >>>>>>>
> >>>>>>>      drivers/leds/Kconfig           |   8 +
> >>>>>>>      drivers/leds/Makefile          |   1 +
> >>>>>>>      drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
> >>>>>>>      3 files changed, 514 insertions(+)
> >>>>>>>      create mode 100644 drivers/leds/leds-is31fl32xx.c
> >>>>>>>
> >>>>>>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> >>>>>>> index 1034696..9c63ba4 100644
> >>>>>>> --- a/drivers/leds/Kconfig
> >>>>>>> +++ b/drivers/leds/Kconfig
> >>>>>>> @@ -580,6 +580,14 @@ config LEDS_SN3218
> >>>>>>>      	  This driver can also be built as a module. If so the module
> >>>>>>>      	  will be called leds-sn3218.
> >>>>>>>
> >>>>>>> +config LEDS_IS31FL32XX
> >>>>>>> +	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
> >>>>>>> +	depends on LEDS_CLASS && I2C && OF
> >>>>>>> +	help
> >>>>>>> +	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
> >>>>>>> +	  They are I2C devices with multiple constant-current channels, each
> >>>>>>> +	  with independent 256-level PWM control.
> >>>>>>> +
> >>>>>>>      comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
> >>>>>>>
> >>>>>>>      config LEDS_BLINKM
> >>>>>>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
> >>>>>>> index 89c9b6f..3fdf313 100644
> >>>>>>> --- a/drivers/leds/Makefile
> >>>>>>> +++ b/drivers/leds/Makefile
> >>>>>>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
> >>>>>>>      obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
> >>>>>>>      obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
> >>>>>>>      obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
> >>>>>>> +obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
> >>>>>>>
> >>>>>>>      # LED SPI Drivers
> >>>>>>>      obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
> >>>>>>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
> >>>>>>> new file mode 100644
> >>>>>>> index 0000000..49818f0
> >>>>>>> --- /dev/null
> >>>>>>> +++ b/drivers/leds/leds-is31fl32xx.c
> >>>>>>> @@ -0,0 +1,505 @@
> >>>>>>> +/*
> >>>>>>> + * linux/drivers/leds-is31fl32xx.c
> >>>>>>> + *
> >>>>>>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
> >>>>>>> + *
> >>>>>>> + * Copyright 2015 Allworx Corp.
> >>>>>>> + *
> >>>>>>> + *
> >>>>>>> + * 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.
> >>>>>>> + *
> >>>>>>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >>>>>>> + */
> >>>>>>> +
> >>>>>>> +#include <linux/err.h>
> >>>>>>> +#include <linux/i2c.h>
> >>>>>>> +#include <linux/kernel.h>
> >>>>>>> +#include <linux/leds.h>
> >>>>>>> +#include <linux/module.h>
> >>>>>>> +#include <linux/of_platform.h>
> >>>>>>> +
> >>>>>>> +/* Used to indicate a device has no such register */
> >>>>>>> +#define IS31FL32XX_REG_NONE 0xFF
> >>>>>>> +
> >>>>>>> +/* Software Shutdown bit in Shutdown Register */
> >>>>>>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
> >>>>>>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
> >>>>>>> +
> >>>>>>> +/* IS31FL3216 has a number of unique registers */
> >>>>>>> +#define IS31FL3216_CONFIG_REG 0x00
> >>>>>>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
> >>>>>>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
> >>>>>>> +
> >>>>>>> +/* Software Shutdown bit in 3216 Config Register */
> >>>>>>> +#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
> >>>>>>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
> >>>>>>> +
> >>>>>>> +struct is31fl32xx_priv;
> >>>>>>> +struct is31fl32xx_led_data {
> >>>>>>> +	struct led_classdev cdev;
> >>>>>>> +	u8 channel; /* 1-based, max priv->cdef->channels */
> >>>>>>> +	struct is31fl32xx_priv *priv;
> >>>>>>> +};
> >>>>>>> +
> >>>>>>> +struct is31fl32xx_priv {
> >>>>>>> +	const struct is31fl32xx_chipdef *cdef;
> >>>>>>> +	struct i2c_client *client;
> >>>>>>> +	unsigned int num_leds;
> >>>>>>> +	struct is31fl32xx_led_data leds[0];
> >>>>>>> +};
> >>>>>>> +
> >>>>>>> +/**
> >>>>>>> + * struct is31fl32xx_chipdef - chip-specific attributes
> >>>>>>> + * @channels            : Number of LED channels
> >>>>>>> + * @shutdown_reg        : address of Shutdown register (optional)
> >>>>>>> + * @pwm_update_reg      : address of PWM Update register
> >>>>>>> + * @global_control_reg  : address of Global Control register (optional)
> >>>>>>> + * @reset_reg           : address of Reset register (optional)
> >>>>>>> + * @pwm_register_base   : address of first PWM register
> >>>>>>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
> >>>>>>> + * @led_control_register_base : address of first LED control register (optional)
> >>>>>>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
> >>>>>>> + * @reset_func:         : pointer to reset function
> >>>>>>> + *
> >>>>>>> + * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
> >>>>>>> + * indicates that this chip has no such register.
> >>>>>>> + *
> >>>>>>> + * If non-NULL, @reset_func will be called during probing to set all
> >>>>>>> + * necessary registers to a known initialization state. This is needed
> >>>>>>> + * for chips that do not have a @reset_reg.
> >>>>>>> + *
> >>>>>>> + * @enable_bits_per_led_control_register must be >=1 if
> >>>>>>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
> >>>>>>> + */
> >>>>>>> +struct is31fl32xx_chipdef {
> >>>>>>> +	u8	channels;
> >>>>>>> +	u8	shutdown_reg;
> >>>>>>> +	u8	pwm_update_reg;
> >>>>>>> +	u8	global_control_reg;
> >>>>>>> +	u8	reset_reg;
> >>>>>>> +	u8	pwm_register_base;
> >>>>>>> +	bool	pwm_registers_reversed;
> >>>>>>> +	u8	led_control_register_base;
> >>>>>>> +	u8	enable_bits_per_led_control_register;
> >>>>>>> +	int (*reset_func)(struct is31fl32xx_priv *priv);
> >>>>>>> +	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
> >>>>>>> +};
> >>>>>>> +
> >>>>>>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
> >>>>>>> +	.channels				= 36,
> >>>>>>> +	.shutdown_reg				= 0x00,
> >>>>>>> +	.pwm_update_reg				= 0x25,
> >>>>>>> +	.global_control_reg			= 0x4a,
> >>>>>>> +	.reset_reg				= 0x4f,
> >>>>>>> +	.pwm_register_base			= 0x01,
> >>>>>>> +	.led_control_register_base		= 0x26,
> >>>>>>> +	.enable_bits_per_led_control_register	= 1,
> >>>>>>> +};
> >>>>>>> +
> >>>>>>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
> >>>>>>> +	.channels				= 28,
> >>>>>>> +	.shutdown_reg				= 0x00,
> >>>>>>> +	.pwm_update_reg				= 0x25,
> >>>>>>> +	.global_control_reg			= 0x4a,
> >>>>>>> +	.reset_reg				= 0x4f,
> >>>>>>> +	.pwm_register_base			= 0x05,
> >>>>>>> +	.led_control_register_base		= 0x2a,
> >>>>>>> +	.enable_bits_per_led_control_register	= 1,
> >>>>>>> +};
> >>>>>>> +
> >>>>>>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
> >>>>>>> +	.channels				= 18,
> >>>>>>> +	.shutdown_reg				= 0x00,
> >>>>>>> +	.pwm_update_reg				= 0x16,
> >>>>>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
> >>>>>>> +	.reset_reg				= 0x17,
> >>>>>>> +	.pwm_register_base			= 0x01,
> >>>>>>> +	.led_control_register_base		= 0x13,
> >>>>>>> +	.enable_bits_per_led_control_register	= 6,
> >>>>>>> +};
> >>>>>>> +
> >>>>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
> >>>>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> >>>>>>> +					bool enable);
> >>>>>>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
> >>>>>>> +	.channels				= 16,
> >>>>>>> +	.shutdown_reg				= IS31FL32XX_REG_NONE,
> >>>>>>> +	.pwm_update_reg				= 0xB0,
> >>>>>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
> >>>>>>> +	.reset_reg				= IS31FL32XX_REG_NONE,
> >>>>>>> +	.pwm_register_base			= 0x10,
> >>>>>>> +	.pwm_registers_reversed			= true,
> >>>>>>> +	.led_control_register_base		= 0x01,
> >>>>>>> +	.enable_bits_per_led_control_register	= 8,
> >>>>>>> +	.reset_func				= is31fl3216_reset,
> >>>>>>> +	.sw_shutdown_func			= is31fl3216_software_shutdown,
> >>>>>>> +};
> >>>>>>> +
> >>>>>>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
> >>>>>>> +{
> >>>>>>> +	int ret;
> >>>>>>> +
> >>>>>>> +	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
> >>>>>>> +
> >>>>>>> +	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
> >>>>>>> +	if (ret) {
> >>>>>>> +		dev_err(&priv->client->dev,
> >>>>>>> +			"register write to 0x%02X failed (error %d)",
> >>>>>>> +			reg, ret);
> >>>>>>> +	}
> >>>>>>> +	return ret;
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +/*
> >>>>>>> + * Custom reset function for IS31FL3216 because it does not have a RESET
> >>>>>>> + * register the way that the other IS31FL32xx chips do. We don't bother
> >>>>>>> + * writing the GPIO and animation registers, because the registers we
> >>>>>>> + * do write ensure those will have no effect.
> >>>>>>> + */
> >>>>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
> >>>>>>> +{
> >>>>>>> +	unsigned int i;
> >>>>>>> +	int ret;
> >>>>>>> +
> >>>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
> >>>>>>> +			       IS31FL3216_CONFIG_SSD_ENABLE);
> >>>>>>> +	if (ret)
> >>>>>>> +		return ret;
> >>>>>>> +	for (i = 0; i < priv->cdef->channels; i++) {
> >>>>>>> +		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
> >>>>>>> +				       0x00);
> >>>>>>> +		if (ret)
> >>>>>>> +			return ret;
> >>>>>>> +	}
> >>>>>>> +	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
> >>>>>>> +	if (ret)
> >>>>>>> +		return ret;
> >>>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
> >>>>>>> +	if (ret)
> >>>>>>> +		return ret;
> >>>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
> >>>>>>> +	if (ret)
> >>>>>>> +		return ret;
> >>>>>>> +
> >>>>>>> +	return 0;
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +/*
> >>>>>>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
> >>>>>>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
> >>>>>>> + * We don't bother doing a read/modify/write on the CONFIG register because
> >>>>>>> + * we only ever use a value of '0' for the other fields in that register.
> >>>>>>> + */
> >>>>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> >>>>>>> +					bool enable)
> >>>>>>> +{
> >>>>>>> +	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
> >>>>>>> +			    IS31FL3216_CONFIG_SSD_DISABLE;
> >>>>>>> +
> >>>>>>> +	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +/*
> >>>>>>> + * NOTE: A mutex is not needed in this function because:
> >>>>>>> + * - All referenced data is read-only after probe()
> >>>>>>> + * - The I2C core has a mutex on to protect the bus
> >>>>>>> + * - There are no read/modify/write operations
> >>>>>>> + * - Intervening operations between the write of the PWM register
> >>>>>>> + *   and the Update register are harmless.
> >>>>>>> + *
> >>>>>>> + * Example:
> >>>>>>> + *	PWM_REG_1 write 16
> >>>>>>> + *	UPDATE_REG write 0
> >>>>>>> + *	PWM_REG_2 write 128
> >>>>>>> + *	UPDATE_REG write 0
> >>>>>>> + *   vs:
> >>>>>>> + *	PWM_REG_1 write 16
> >>>>>>> + *	PWM_REG_2 write 128
> >>>>>>> + *	UPDATE_REG write 0
> >>>>>>> + *	UPDATE_REG write 0
> >>>>>>> + * are equivalent. Poking the Update register merely applies all PWM
> >>>>>>> + * register writes up to that point.
> >>>>>>> + */
> >>>>>>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
> >>>>>>> +				     enum led_brightness brightness)
> >>>>>>> +{
> >>>>>>> +	const struct is31fl32xx_led_data *led_data =
> >>>>>>> +		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
> >>>>>>> +	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
> >>>>>>> +	u8 pwm_register_offset;
> >>>>>>> +	int ret;
> >>>>>>> +
> >>>>>>> +	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
> >>>>>>> +
> >>>>>>> +	/* NOTE: led_data->channel is 1-based */
> >>>>>>> +	if (cdef->pwm_registers_reversed)
> >>>>>>> +		pwm_register_offset = cdef->channels - led_data->channel;
> >>>>>>> +	else
> >>>>>>> +		pwm_register_offset = led_data->channel - 1;
> >>>>>>> +
> >>>>>>> +	ret = is31fl32xx_write(led_data->priv,
> >>>>>>> +			       cdef->pwm_register_base + pwm_register_offset,
> >>>>>>> +			       brightness);
> >>>>>>> +	if (ret)
> >>>>>>> +		return ret;
> >>>>>>> +
> >>>>>>> +	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
> >>>>>>> +{
> >>>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>>>>>> +	int ret;
> >>>>>>> +
> >>>>>>> +	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
> >>>>>>> +		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
> >>>>>>> +		if (ret)
> >>>>>>> +			return ret;
> >>>>>>> +	}
> >>>>>>> +
> >>>>>>> +	if (cdef->reset_func)
> >>>>>>> +		return cdef->reset_func(priv);
> >>>>>>> +
> >>>>>>> +	return 0;
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
> >>>>>>> +					bool enable)
> >>>>>>> +{
> >>>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>>>>>> +	int ret;
> >>>>>>> +
> >>>>>>> +	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
> >>>>>>> +		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
> >>>>>>> +				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
> >>>>>>> +		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
> >>>>>>> +		if (ret)
> >>>>>>> +			return ret;
> >>>>>>> +	}
> >>>>>>> +
> >>>>>>> +	if (cdef->sw_shutdown_func)
> >>>>>>> +		return cdef->sw_shutdown_func(priv, enable);  
> >>>>>>
> >>>>>> You seem to call sw_shutdown_func only here, so why should we have
> >>>>>> enable parameter in this op?  
> >>>>>
> >>>>> I'm not sure if I understand the question, but I will try to answer.
> >>>>>
> >>>>> 'enable' is passed through is31fl32xx_software_shutdown to
> >>>>> cdef->sw_shutdown_func, so it can be either true or false at that
> >>>>> point. The purpose of sw_shutdown_func is to add any special behavior
> >>>>> when enabling/disabling software-shutdown mode, which is needed for
> >>>>> the 3216 because its SSD bit is in a different position and with
> >>>>> opposite polarity.
> >>>>>
> >>>>> Is it that 'enable' in that line of code makes it look like it's being
> >>>>> called with an hardcoded value rather than a variable? If so, perhaps a
> >>>>> different parameter name would make it more obvious? Or a kerneldoc
> >>>>> comment for the function to describe the parameter?
> >>>>>
> >>>>> Or have I totally missed the point of the question?  
> >>>>
> >>>> Actually I should have placed this question next to
> >>>> the call to s31fl32xx_software_shutdown() in is31fl32xx_init_regs(),
> >>>> which is passed "false" in the second argument, and there is no
> >>>> other call to s31fl32xx_software_shutdown() in the driver.
> >>>>
> >>>> Having the argument makes people wondering that there is some
> >>>> use case in the driver, where "true" is passed, but it seems not
> >>>> to be the case.  
> >>>
> >>> Thanks for the clarification.
> >>>
> >>> Yes, there is currently no explicit call to enable software-shutdown
> >>> mode. Since the reset state of these devices is to have software-shutdown
> >>> enabled, the only explicit operation on it that's currently needed is
> >>> to disable it.
> >>>
> >>> Reasons I can think to have the 'enable' parameter anyways include:
> >>> - It seems logical to me that an API to manipulate the software-shutdown
> >>>     state should allow both enabling and disabling.
> >>> - Having a parameter for that seemed the most obvious way to go, and it
> >>>     was trivial to implement.
> >>> - Alternatively having separate "enable" and "disable" functions would
> >>>     duplicate most of the logic, vs the parameter being just a single
> >>>     conditional. And that would also imply two function pointers in the
> >>>     chipdefs, which I'd prefer to minimize.
> >>> - If anyone wanted to implement suspend/resume in the future, they would
> >>>     most likely do it by enabling/disabling software-shutdown. Supporting
> >>>     both enable/disable from the start should make that trivial to do.  
> >>
> >> Suspend/resume callbacks are already implemented in led-class.c and
> >> related ops indirectly call brightness_set. If you want to support
> >> suspend/resume you have to set LED_CORE_SUSPENDRESUME flag.  
> >
> > If I understand correctly, all LED_CORE_SUSPENDRESUME will do is cause
> > the led core to set the brightness to 0 on suspend (and reverse that
> > on resume). I see some drivers use this flag and other do not.
> > This brings up the question in my mind: how would a driver decide
> > whether it is  appropriate for an LED to go dark on suspend? Is that
> > just the drivers that know the logical purpose of the LEDs they control?  
> 
> There is a room for improvement here. Possibly a new LED class sysfs
> attribute could be of help in determining that.
> 
> >> Setting brightness to 0 is equivalent to turning the LED controller
> >> in a shutdown mode, provided that all sub-LEDs are off.  
> >
> > This is not strictly true for these devices. If someone cared very much
> > about power usage when suspended they may want to put the LED controller
> > into what it calls "shutdown mode" via the SSD bit. I didn't bother mainly
> > because I have no need, and also because I wasn't sure how to even trigger
> > a suspend in order to test an implementation.  
> 
> I meant that LED class driver should put the device in a software
> shutdown mode after last sub-LED is turned off.
> 
> > FYI, I just measured it the effect of software-shutdown even with all LEDs
> > already off. The difference in current is about 5mA, measured at the 5V
> > supply to a 3216 eval board. Not huge, but someone might care.  
> 
> This can be vital difference for some use cases. You could count the
> number of currently active sub-LEDs and put the controller in a software
> shutdown mode in case the value is 0.

I had thought about that, but I had some concerns that dissuaded me from
perusing it:
- Would need some way to know whether a brightness_set has changed an LED 
  from on->off or off->on in order to update that counter. leds-core
  modifies led_cdev->brightness before calling brightness_set, so that
  information would need to be somehow maintained in the driver.
- It requires an additional mutex to protect whatever data is involved in 
  the previous bullet. That mutex would need to be taken on every brightness 
  update.
- If a single LED is blinking rapidly (e.g. a cpu trigger), it would thrash 
  software shutdown mode. That in turn adds even more CPU usage (due to
  register writes and mutexes), potentially making power usage worse in 
  some cases.
- Time it might take to implement and review (mostly depending on how the
  first bullet is handled), especially since the 4.6 merge window is fast
  approaching. 

That said, I could propose the following simple-to-implement design:
- counter, and mutex to protect it, added to 'priv'
- previous_brightness field added to 'led_data' 
- brightness_set compares new brightness vs previous
   - if on/off state has changed: 
     - takes mutex 
     - updates counter
     - sets software-shutdown mode to (counter==0)
     - releases mutex

If you think that's worth doing, I can try to get it in along with the
other changes for v2. 

> >>> - I thought the code read better with a bool parameter, vs a longer
> >>>     function name.
> >>>
> >>> So nothing really critical, but mostly just my aesthetic and preference.
> >>>
> >>> Also, I expect that is31fl32xx_software_shutdown() would be inlined, so
> >>> the conditional check in there is optimized out anyways, and there is no
> >>> performance penalty. Looking at the disassembly in my ARMv7a build, both
> >>> of those have indeed happened there.  
> >>
> >> Not only the size of a binary and the performance, but also code
> >> readability matters. The function is called only with false argument,
> >> so I expect that some people may submit patches optimizing this.
> >>
> >> Let's avoid the confusion.  
> >
> > I guess we just have a difference of opinion on which way is more
> > readable, which is OK. Unless the above explanation causes you to
> > change your mind, I will remove the 'enable' parameter and add a
> > "_disable" suffix to both functions. That will also leave the
> > 	#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
> > constant unused in the code, should that also be removed? Note that
> > the 3216 specific constant
> > 	#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
> > would still be used in is31fl3216_reset().  
> 
> Current version of the function would be required for enabling
> shutting down the controller from brightness_set op when the
> count of active sub-LEDs drops to 0.

Agreed. That's a reason I would have for leaving it so even if the
auto-software-shutdown logic is not implemented at this time. But I 
would understand if you still prefer to only have the current form 
if there is currently code which uses both possible values of the 
'enable' parameter.

> >>>>>>> +	return 0;
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
> >>>>>>> +{
> >>>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>>>>>> +	int ret;
> >>>>>>> +
> >>>>>>> +	ret = is31fl32xx_reset_regs(priv);
> >>>>>>> +	if (ret)
> >>>>>>> +		return ret;
> >>>>>>> +
> >>>>>>> +	/*
> >>>>>>> +	 * Set enable bit for all channels.
> >>>>>>> +	 * We will control state with PWM registers alone.
> >>>>>>> +	 */
> >>>>>>> +	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
> >>>>>>> +		u8 value =
> >>>>>>> +		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
> >>>>>>> +		u8 num_regs = cdef->channels /
> >>>>>>> +				cdef->enable_bits_per_led_control_register;
> >>>>>>> +		int i;
> >>>>>>> +
> >>>>>>> +		for (i = 0; i < num_regs; i++) {
> >>>>>>> +			ret = is31fl32xx_write(priv,
> >>>>>>> +					       cdef->led_control_register_base+i,
> >>>>>>> +					       value);
> >>>>>>> +			if (ret)
> >>>>>>> +				return ret;
> >>>>>>> +		}
> >>>>>>> +	}
> >>>>>>> +
> >>>>>>> +	ret = is31fl32xx_software_shutdown(priv, false);
> >>>>>>> +	if (ret)
> >>>>>>> +		return ret;
> >>>>>>> +
> >>>>>>> +	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
> >>>>>>> +		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
> >>>>>>> +		if (ret)
> >>>>>>> +			return ret;
> >>>>>>> +	}
> >>>>>>> +
> >>>>>>> +	return 0;
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
> >>>>>>> +{
> >>>>>>> +	return sizeof(struct is31fl32xx_priv) +
> >>>>>>> +		      (sizeof(struct is31fl32xx_led_data) * num_leds);
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
> >>>>>>> +				     const struct device_node *child,
> >>>>>>> +				     struct is31fl32xx_led_data *led_data)
> >>>>>>> +{
> >>>>>>> +	struct led_classdev *cdev = &led_data->cdev;
> >>>>>>> +	int ret = 0;
> >>>>>>> +	u32 reg;
> >>>>>>> +
> >>>>>>> +	if (of_property_read_string(child, "label", &cdev->name))
> >>>>>>> +		cdev->name = child->name;
> >>>>>>> +
> >>>>>>> +	ret = of_property_read_u32(child, "reg", &reg);
> >>>>>>> +	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
> >>>>>>> +		dev_err(dev,
> >>>>>>> +			"Child node %s does not have a valid reg property\n",
> >>>>>>> +			child->full_name);
> >>>>>>> +		return -EINVAL;
> >>>>>>> +	}
> >>>>>>> +	led_data->channel = reg;
> >>>>>>> +
> >>>>>>> +	of_property_read_string(child, "linux,default-trigger",
> >>>>>>> +				&cdev->default_trigger);
> >>>>>>> +
> >>>>>>> +	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
> >>>>>>> +
> >>>>>>> +	return 0;
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
> >>>>>>> +					struct is31fl32xx_priv *priv,
> >>>>>>> +					u8 channel)
> >>>>>>> +{
> >>>>>>> +	size_t i;
> >>>>>>> +
> >>>>>>> +	for (i = 0; i < priv->num_leds; i++) {
> >>>>>>> +		if (priv->leds[i].channel == channel)
> >>>>>>> +			return &priv->leds[i];
> >>>>>>> +	}
> >>>>>>> +
> >>>>>>> +	return NULL;
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +static int is31fl32xx_parse_dt(struct device *dev,
> >>>>>>> +			       struct is31fl32xx_priv *priv)
> >>>>>>> +{
> >>>>>>> +	struct device_node *child;
> >>>>>>> +	int ret = 0;
> >>>>>>> +
> >>>>>>> +	for_each_child_of_node(dev->of_node, child) {
> >>>>>>> +		struct is31fl32xx_led_data *led_data =
> >>>>>>> +			&priv->leds[priv->num_leds];
> >>>>>>> +		const struct is31fl32xx_led_data *other_led_data;
> >>>>>>> +
> >>>>>>> +		led_data->priv = priv;
> >>>>>>> +
> >>>>>>> +		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
> >>>>>>> +		if (ret)
> >>>>>>> +			goto err;
> >>>>>>> +
> >>>>>>> +		/* Detect if channel is already in use by another child */
> >>>>>>> +		other_led_data = is31fl32xx_find_led_data(priv,
> >>>>>>> +							  led_data->channel);
> >>>>>>> +		if (other_led_data) {
> >>>>>>> +			dev_err(dev,
> >>>>>>> +				"%s and %s both attempting to use channel %d\n",
> >>>>>>> +				led_data->cdev.name,
> >>>>>>> +				other_led_data->cdev.name,
> >>>>>>> +				led_data->channel);
> >>>>>>> +			goto err;
> >>>>>>> +		}
> >>>>>>> +
> >>>>>>> +		ret = devm_led_classdev_register(dev, &led_data->cdev);
> >>>>>>> +		if (ret) {
> >>>>>>> +			dev_err(dev, "failed to register PWM led for %s: %d\n",
> >>>>>>> +				led_data->cdev.name, ret);
> >>>>>>> +			goto err;
> >>>>>>> +		}
> >>>>>>> +
> >>>>>>> +		priv->num_leds++;
> >>>>>>> +	}
> >>>>>>> +
> >>>>>>> +	return 0;
> >>>>>>> +
> >>>>>>> +err:
> >>>>>>> +	of_node_put(child);
> >>>>>>> +	return ret;
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +static const struct of_device_id of_is31fl31xx_match[] = {
> >>>>>>> +	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
> >>>>>>> +	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
> >>>>>>> +	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
> >>>>>>> +	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
> >>>>>>> +	{},
> >>>>>>> +};
> >>>>>>> +
> >>>>>>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
> >>>>>>> +
> >>>>>>> +static int is31fl32xx_probe(struct i2c_client *client,
> >>>>>>> +			    const struct i2c_device_id *id)
> >>>>>>> +{
> >>>>>>> +	const struct is31fl32xx_chipdef *cdef;
> >>>>>>> +	const struct of_device_id *of_dev_id;
> >>>>>>> +	struct device *dev = &client->dev;
> >>>>>>> +	struct is31fl32xx_priv *priv;
> >>>>>>> +	int count;
> >>>>>>> +	int ret = 0;
> >>>>>>> +
> >>>>>>> +	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
> >>>>>>> +	if (!of_dev_id)
> >>>>>>> +		return -EINVAL;
> >>>>>>> +
> >>>>>>> +	cdef = of_dev_id->data;
> >>>>>>> +
> >>>>>>> +	count = of_get_child_count(dev->of_node);
> >>>>>>> +	if (!count)
> >>>>>>> +		return -EINVAL;
> >>>>>>> +
> >>>>>>> +	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
> >>>>>>> +			    GFP_KERNEL);
> >>>>>>> +	if (!priv)
> >>>>>>> +		return -ENOMEM;
> >>>>>>> +
> >>>>>>> +	priv->client = client;
> >>>>>>> +	priv->cdef = cdef;
> >>>>>>> +	i2c_set_clientdata(client, priv);
> >>>>>>> +
> >>>>>>> +	ret = is31fl32xx_init_regs(priv);
> >>>>>>> +	if (ret)
> >>>>>>> +		return ret;
> >>>>>>> +
> >>>>>>> +	ret = is31fl32xx_parse_dt(dev, priv);
> >>>>>>> +	if (ret)
> >>>>>>> +		return ret;
> >>>>>>> +
> >>>>>>> +	return 0;
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +static int is31fl32xx_remove(struct i2c_client *client)
> >>>>>>> +{
> >>>>>>> +	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
> >>>>>>> +
> >>>>>>> +	return is31fl32xx_reset_regs(priv);
> >>>>>>> +}
> >>>>>>> +
> >>>>>>> +/*
> >>>>>>> + * i2c-core requires that id_table be non-NULL, even though
> >>>>>>> + * it is not used for DeviceTree based instantiation.
> >>>>>>> + */
> >>>>>>> +static const struct i2c_device_id is31fl31xx_id[] = {
> >>>>>>> +	{},
> >>>>>>> +};
> >>>>>>> +
> >>>>>>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
> >>>>>>> +
> >>>>>>> +static struct i2c_driver is31fl32xx_driver = {
> >>>>>>> +	.driver = {
> >>>>>>> +		.name	= "is31fl32xx",
> >>>>>>> +		.of_match_table = of_is31fl31xx_match,
> >>>>>>> +	},
> >>>>>>> +	.probe		= is31fl32xx_probe,
> >>>>>>> +	.remove		= is31fl32xx_remove,
> >>>>>>> +	.id_table	= is31fl31xx_id,
> >>>>>>> +};
> >>>>>>> +
> >>>>>>> +module_i2c_driver(is31fl32xx_driver);
> >>>>>>> +
> >>>>>>> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
> >>>>>>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
> >>>>>>> +MODULE_LICENSE("GPL v2");
> >>>>>>>  

Re: [PATCH 4/4] leds: Replace dedicated SN3218 driver with IS31FL32XX driver

[David Rivshin (Allworx)]

On Thu, 03 Mar 2016 15:51:36 +0100
Jacek Anaszewski <j.anaszewski@samsung.com> wrote:

> Hi David,
> 
> I'll wait for Tested-by from Stefan before applying this patch.
> If Stefan will have managed to test your driver with his hardware
> by the end of this cycle, it will suffice for this patch to contain
> only leds-is31fl32xx extension part.
> 
> leds-sn3218 hasn't been merged to mainline yet, so I'd rather
> remove it when I get Tested-by from Stefan.
> 
> Otherwise, I will send leds-sn3218 to Linus, and this patch
> will be applied after being tested, during 4.7 cycle.

Since Stefan has given his Tested-by, do I understand correctly
that you would prefer to:
 - revert sn3218 patches 2 and 3 yourself
 - have me provide v2 patches that would apply ontop of that

If so, should I do as Rob suggests and fold the sn3216/3218 bits into 
the earlier patches, or leave it as a 4th patch? I'd probably prefer 
the former as being an easier workflow (less "rebase -i"ing).

And should I base off v4.5-rc6 at that point, or your for-next minus 
those patches? The only difference should be a 1-line offset in the 
vendor-prefixes.txt hunk. Either way is equally easy for me.

> 
> Thanks,
> Jacek Anaszewski
> 
> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:
> > From: David Rivshin <drivshin@allworx.com>
> >
> > Si-En Technology was acquired by ISSI in 2011, and it appears that
> > the IS31FL3218/IS31FL3216 are just rebranded SN3218/SN3216 devices.
> > As the IS31FL32XX driver already handles the *3218 devices, there
> > is no longer a need for the dedicated SN3218 driver.
> >
> > Add the "sn,sn3218" and "sn,sn3216" compatible strings into the
> > IS31FL32XX driver and binding documentation, and remove the
> > leds-sn3218 driver.
> >
> > Datasheets:
> >      IS31FL3218: http://www.issi.com/WW/pdf/31FL3218.pdf
> >      SN3218:     http://www.si-en.com/uploadpdf/s2011517171720.pdf
> >
> >      IS31FL3216: http://www.issi.com/WW/pdf/31FL3216.pdf
> >      SN3216;     http://www.si-en.com/uploadpdf/SN3216201152410148.pdf
> >
> > Signed-off-by: David Rivshin <drivshin@allworx.com>
> > ---
> >
> > Note that the leds-sn3218 binding use a 0-based 'reg' property, while
> > the leds-is31fl32xx binding uses a 1-based 'reg' property. This seemed
> > to be the preferred binding based on [1]. Since leds-sn3216 has not been
> > in a released kernel, there is are no backwards-compatibility concerns.
> >
> > Changes from RFC:
> >   new
> >
> > [1] http://www.spinics.net/lists/linux-leds/msg05589.html
> >
> >   .../devicetree/bindings/leds/leds-is31fl32xx.txt   |   9 +-
> >   .../devicetree/bindings/leds/leds-sn3218.txt       |  41 ---
> >   drivers/leds/Kconfig                               |  18 +-
> >   drivers/leds/Makefile                              |   1 -
> >   drivers/leds/leds-is31fl32xx.c                     |   6 +-
> >   drivers/leds/leds-sn3218.c                         | 306 ---------------------
> >   6 files changed, 14 insertions(+), 367 deletions(-)
> >   delete mode 100644 Documentation/devicetree/bindings/leds/leds-sn3218.txt
> >   delete mode 100644 drivers/leds/leds-sn3218.c
> >
> > diff --git a/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
> > index 539df2e..c59eb1a 100644
> > --- a/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
> > +++ b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
> > @@ -1,6 +1,6 @@
> > -Binding for ISSI IS31FL32xx LED Drivers
> > +Binding for ISSI IS31FL32xx and Si-En SN32xx LED Drivers
> >
> > -The IS31FL32xx family of LED drivers are I2C devices with multiple
> > +The IS31FL32xx/SN32xx family of LED drivers are I2C devices with multiple
> >   constant-current channels, each with independent 256-level PWM control.
> >   Each LED is represented as a sub-node of the device.
> >
> > @@ -10,6 +10,8 @@ Required properties:
> >   	issi,is31fl3235
> >   	issi,is31fl3218
> >   	issi,is31fl3216
> > +	si-en,sn3218
> > +	si-en,sn3216
> >   - reg: I2C slave address
> >   - address-cells : must be 1
> >   - size-cells : must be 0
> > @@ -45,5 +47,6 @@ leds: is31fl3236@3c {
> >   	};
> >   };
> >
> > -For more product information please see the link below:
> > +For more product information please see the links below:
> >   http://www.issi.com/US/product-analog-fxled-driver.shtml
> > +http://www.si-en.com/product.asp?parentid=890
> > diff --git a/Documentation/devicetree/bindings/leds/leds-sn3218.txt b/Documentation/devicetree/bindings/leds/leds-sn3218.txt
> > deleted file mode 100644
> > index 19cbf57..0000000
> > --- a/Documentation/devicetree/bindings/leds/leds-sn3218.txt
> > +++ /dev/null
> > @@ -1,41 +0,0 @@
> > -* Si-En Technology - SN3218 18-Channel LED Driver
> > -
> > -Required properties:
> > -- compatible :
> > -	"si-en,sn3218"
> > -- address-cells : must be 1
> > -- size-cells : must be 0
> > -- reg : I2C slave address, typically 0x54
> > -
> > -There must be at least 1 LED which is represented as a sub-node
> > -of the sn3218 device.
> > -
> > -LED sub-node properties:
> > -- label : (optional) see Documentation/devicetree/bindings/leds/common.txt
> > -- reg : number of LED line (could be from 0 to 17)
> > -- linux,default-trigger : (optional)
> > -   see Documentation/devicetree/bindings/leds/common.txt
> > -
> > -Example:
> > -
> > -sn3218: led-controller@54 {
> > -	compatible = "si-en,sn3218";
> > -	#address-cells = <1>;
> > -	#size-cells = <0>;
> > -	reg = <0x54>;
> > -
> > -	led@0 {
> > -		label = "led1";
> > -		reg = <0x0>;
> > -	};
> > -
> > -	led@1 {
> > -		label = "led2";
> > -		reg = <0x1>;
> > -	};
> > -
> > -	led@2 {
> > -		label = "led3";
> > -		reg = <0x2>;
> > -	};
> > -};
> > diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> > index 9c63ba4..1f64151 100644
> > --- a/drivers/leds/Kconfig
> > +++ b/drivers/leds/Kconfig
> > @@ -568,25 +568,13 @@ config LEDS_SEAD3
> >   	  This driver can also be built as a module. If so the module
> >   	  will be called leds-sead3.
> >
> > -config LEDS_SN3218
> > -	tristate "LED support for Si-En SN3218 I2C chip"
> > -	depends on LEDS_CLASS && I2C
> > -	depends on OF
> > -	select REGMAP_I2C
> > -	help
> > -	  This option enables support for the Si-EN SN3218 LED driver
> > -	  connected through I2C. Say Y to enable support for the SN3218 LED.
> > -
> > -	  This driver can also be built as a module. If so the module
> > -	  will be called leds-sn3218.
> > -
> >   config LEDS_IS31FL32XX
> >   	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
> >   	depends on LEDS_CLASS && I2C && OF
> >   	help
> > -	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
> > -	  They are I2C devices with multiple constant-current channels, each
> > -	  with independent 256-level PWM control.
> > +	  Say Y here to include support for ISSI IS31FL32XX and Si-En SN32xx
> > +	  LED controllers. They are I2C devices with multiple constant-current
> > +	  channels, each with independent 256-level PWM control.
> >
> >   comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
> >
> > diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
> > index 3fdf313..cb2013d 100644
> > --- a/drivers/leds/Makefile
> > +++ b/drivers/leds/Makefile
> > @@ -66,7 +66,6 @@ obj-$(CONFIG_LEDS_MENF21BMC)		+= leds-menf21bmc.o
> >   obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
> >   obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
> >   obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
> > -obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
> >   obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
> >
> >   # LED SPI Drivers
> > diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
> > index 49818f0..ec3f541 100644
> > --- a/drivers/leds/leds-is31fl32xx.c
> > +++ b/drivers/leds/leds-is31fl32xx.c
> > @@ -10,7 +10,9 @@
> >    * it under the terms of the GNU General Public License version 2 as
> >    * published by the Free Software Foundation.
> >    *
> > - * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> > + * Datasheets:
> > + *   http://www.issi.com/US/product-analog-fxled-driver.shtml
> > + *   http://www.si-en.com/product.asp?parentid=890
> >    */
> >
> >   #include <linux/err.h>
> > @@ -425,7 +427,9 @@ static const struct of_device_id of_is31fl31xx_match[] = {
> >   	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
> >   	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
> >   	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
> > +	{ .compatible = "si-en,sn3218",    .data = &is31fl3218_cdef, },
> >   	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
> > +	{ .compatible = "si-en,sn3216",    .data = &is31fl3216_cdef, },
> >   	{},
> >   };
> >
> > diff --git a/drivers/leds/leds-sn3218.c b/drivers/leds/leds-sn3218.c
> > deleted file mode 100644
> > index dcc2581..0000000
> > --- a/drivers/leds/leds-sn3218.c
> > +++ /dev/null
> > @@ -1,306 +0,0 @@
> > -/*
> > - * Si-En SN3218 18 Channel LED Driver
> > - *
> > - * Copyright (C) 2016 Stefan Wahren <stefan.wahren@i2se.com>
> > - *
> > - * Based on leds-pca963x.c
> > - *
> > - * 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.
> > - *
> > - * Datasheet: http://www.si-en.com/uploadpdf/s2011517171720.pdf
> > - *
> > - */
> > -
> > -#include <linux/err.h>
> > -#include <linux/i2c.h>
> > -#include <linux/leds.h>
> > -#include <linux/module.h>
> > -#include <linux/of.h>
> > -#include <linux/regmap.h>
> > -#include <linux/slab.h>
> > -
> > -#define SN3218_MODE		0x00
> > -#define SN3218_PWM_1		0x01
> > -#define SN3218_PWM_2		0x02
> > -#define SN3218_PWM_3		0x03
> > -#define SN3218_PWM_4		0x04
> > -#define SN3218_PWM_5		0x05
> > -#define SN3218_PWM_6		0x06
> > -#define SN3218_PWM_7		0x07
> > -#define SN3218_PWM_8		0x08
> > -#define SN3218_PWM_9		0x09
> > -#define SN3218_PWM_10		0x0a
> > -#define SN3218_PWM_11		0x0b
> > -#define SN3218_PWM_12		0x0c
> > -#define SN3218_PWM_13		0x0d
> > -#define SN3218_PWM_14		0x0e
> > -#define SN3218_PWM_15		0x0f
> > -#define SN3218_PWM_16		0x10
> > -#define SN3218_PWM_17		0x11
> > -#define SN3218_PWM_18		0x12
> > -#define SN3218_LED_1_6		0x13
> > -#define SN3218_LED_7_12		0x14
> > -#define SN3218_LED_13_18	0x15
> > -#define SN3218_UPDATE		0x16	/* applies to reg 0x01 .. 0x15 */
> > -#define SN3218_RESET		0x17
> > -
> > -#define SN3218_LED_MASK		0x3f
> > -#define SN3218_LED_ON		0x01
> > -#define SN3218_LED_OFF		0x00
> > -
> > -#define SN3218_MODE_SHUTDOWN	0x00
> > -#define SN3218_MODE_NORMAL	0x01
> > -
> > -#define NUM_LEDS		18
> > -
> > -struct sn3218_led;
> > -
> > -/**
> > - * struct sn3218 -
> > - * @client - Pointer to the I2C client
> > - * @leds - Pointer to the individual LEDs
> > - * @num_leds - Actual number of LEDs
> > -**/
> > -struct sn3218 {
> > -	struct i2c_client *client;
> > -	struct regmap *regmap;
> > -	struct sn3218_led *leds;
> > -	int num_leds;
> > -};
> > -
> > -/**
> > - * struct sn3218_led -
> > - * @chip - Pointer to the container
> > - * @led_cdev - led class device pointer
> > - * @led_num - LED index ( 0 .. 17 )
> > -**/
> > -struct sn3218_led {
> > -	struct sn3218 *chip;
> > -	struct led_classdev led_cdev;
> > -	int led_num;
> > -};
> > -
> > -static int sn3218_led_set(struct led_classdev *led_cdev,
> > -			  enum led_brightness brightness)
> > -{
> > -	struct sn3218_led *led =
> > -			container_of(led_cdev, struct sn3218_led, led_cdev);
> > -	struct regmap *regmap = led->chip->regmap;
> > -	u8 bank = led->led_num / 6;
> > -	u8 mask = 0x1 << (led->led_num % 6);
> > -	u8 val;
> > -	int ret;
> > -
> > -	if (brightness == LED_OFF)
> > -		val = 0;
> > -	else
> > -		val = mask;
> > -
> > -	ret = regmap_update_bits(regmap, SN3218_LED_1_6 + bank, mask, val);
> > -	if (ret < 0)
> > -		return ret;
> > -
> > -	if (brightness > LED_OFF) {
> > -		ret = regmap_write(regmap, SN3218_PWM_1 + led->led_num,
> > -				   brightness);
> > -		if (ret < 0)
> > -			return ret;
> > -	}
> > -
> > -	ret = regmap_write(regmap, SN3218_UPDATE, 0xff);
> > -
> > -	return ret;
> > -}
> > -
> > -static void sn3218_led_init(struct sn3218 *sn3218, struct device_node *node,
> > -			    u32 reg)
> > -{
> > -	struct sn3218_led *leds = sn3218->leds;
> > -	struct led_classdev *cdev = &leds[reg].led_cdev;
> > -
> > -	leds[reg].led_num = reg;
> > -	leds[reg].chip = sn3218;
> > -
> > -	if (of_property_read_string(node, "label", &cdev->name))
> > -		cdev->name = node->name;
> > -
> > -	of_property_read_string(node, "linux,default-trigger",
> > -				&cdev->default_trigger);
> > -
> > -	cdev->brightness_set_blocking = sn3218_led_set;
> > -}
> > -
> > -static const struct reg_default sn3218_reg_defs[] = {
> > -	{ SN3218_MODE, 0x00},
> > -	{ SN3218_PWM_1, 0x00},
> > -	{ SN3218_PWM_2, 0x00},
> > -	{ SN3218_PWM_3, 0x00},
> > -	{ SN3218_PWM_4, 0x00},
> > -	{ SN3218_PWM_5, 0x00},
> > -	{ SN3218_PWM_6, 0x00},
> > -	{ SN3218_PWM_7, 0x00},
> > -	{ SN3218_PWM_8, 0x00},
> > -	{ SN3218_PWM_9, 0x00},
> > -	{ SN3218_PWM_10, 0x00},
> > -	{ SN3218_PWM_11, 0x00},
> > -	{ SN3218_PWM_12, 0x00},
> > -	{ SN3218_PWM_13, 0x00},
> > -	{ SN3218_PWM_14, 0x00},
> > -	{ SN3218_PWM_15, 0x00},
> > -	{ SN3218_PWM_16, 0x00},
> > -	{ SN3218_PWM_17, 0x00},
> > -	{ SN3218_PWM_18, 0x00},
> > -	{ SN3218_LED_1_6, 0x00},
> > -	{ SN3218_LED_7_12, 0x00},
> > -	{ SN3218_LED_13_18, 0x00},
> > -	{ SN3218_UPDATE, 0x00},
> > -	{ SN3218_RESET, 0x00},
> > -};
> > -
> > -static const struct regmap_config sn3218_regmap_config = {
> > -	.reg_bits = 8,
> > -	.val_bits = 8,
> > -
> > -	.max_register = SN3218_RESET,
> > -	.reg_defaults = sn3218_reg_defs,
> > -	.num_reg_defaults = ARRAY_SIZE(sn3218_reg_defs),
> > -	.cache_type = REGCACHE_RBTREE,
> > -};
> > -
> > -static int sn3218_init(struct i2c_client *client, struct sn3218 *sn3218)
> > -{
> > -	struct device_node *np = client->dev.of_node, *child;
> > -	struct sn3218_led *leds;
> > -	int ret, count;
> > -
> > -	count = of_get_child_count(np);
> > -	if (!count)
> > -		return -ENODEV;
> > -
> > -	if (count > NUM_LEDS) {
> > -		dev_err(&client->dev, "Invalid LED count %d\n", count);
> > -		return -EINVAL;
> > -	}
> > -
> > -	leds = devm_kzalloc(&client->dev, count * sizeof(*leds), GFP_KERNEL);
> > -	if (!leds)
> > -		return -ENOMEM;
> > -
> > -	sn3218->leds = leds;
> > -	sn3218->num_leds = count;
> > -	sn3218->client = client;
> > -
> > -	sn3218->regmap = devm_regmap_init_i2c(client, &sn3218_regmap_config);
> > -	if (IS_ERR(sn3218->regmap)) {
> > -		ret = PTR_ERR(sn3218->regmap);
> > -		dev_err(&client->dev, "Failed to allocate register map: %d\n",
> > -			ret);
> > -		return ret;
> > -	}
> > -
> > -	for_each_child_of_node(np, child) {
> > -		u32 reg;
> > -
> > -		ret = of_property_read_u32(child, "reg", &reg);
> > -		if (ret)
> > -			goto fail;
> > -
> > -		if (reg >= count) {
> > -			dev_err(&client->dev, "Invalid LED (%u >= %d)\n", reg,
> > -				count);
> > -			ret = -EINVAL;
> > -			goto fail;
> > -		}
> > -
> > -		sn3218_led_init(sn3218, child, reg);
> > -	}
> > -
> > -	return 0;
> > -
> > -fail:
> > -	of_node_put(child);
> > -	return ret;
> > -}
> > -
> > -static int sn3218_probe(struct i2c_client *client,
> > -			const struct i2c_device_id *id)
> > -{
> > -	struct sn3218 *sn3218;
> > -	struct sn3218_led *leds;
> > -	struct device *dev = &client->dev;
> > -	int i, ret;
> > -
> > -	sn3218 = devm_kzalloc(dev, sizeof(*sn3218), GFP_KERNEL);
> > -	if (!sn3218)
> > -		return -ENOMEM;
> > -
> > -	ret = sn3218_init(client, sn3218);
> > -	if (ret)
> > -		return ret;
> > -
> > -	i2c_set_clientdata(client, sn3218);
> > -	leds = sn3218->leds;
> > -
> > -	/*
> > -	 * Since the chip is write-only we need to reset him into
> > -	 * a defined state (all LEDs off).
> > -	 */
> > -	ret = regmap_write(sn3218->regmap, SN3218_RESET, 0xff);
> > -	if (ret)
> > -		return ret;
> > -
> > -	for (i = 0; i < sn3218->num_leds; i++) {
> > -		ret = devm_led_classdev_register(dev, &leds[i].led_cdev);
> > -		if (ret < 0)
> > -			return ret;
> > -	}
> > -
> > -	return regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_NORMAL);
> > -}
> > -
> > -static int sn3218_remove(struct i2c_client *client)
> > -{
> > -	struct sn3218 *sn3218 = i2c_get_clientdata(client);
> > -
> > -	regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_SHUTDOWN);
> > -
> > -	return 0;
> > -}
> > -
> > -static void sn3218_shutdown(struct i2c_client *client)
> > -{
> > -	struct sn3218 *sn3218 = i2c_get_clientdata(client);
> > -
> > -	regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_SHUTDOWN);
> > -}
> > -
> > -static const struct i2c_device_id sn3218_id[] = {
> > -	{ "sn3218", 0 },
> > -	{ }
> > -};
> > -MODULE_DEVICE_TABLE(i2c, sn3218_id);
> > -
> > -static const struct of_device_id of_sn3218_match[] = {
> > -	{ .compatible = "si-en,sn3218", },
> > -	{},
> > -};
> > -MODULE_DEVICE_TABLE(of, of_sn3218_match);
> > -
> > -static struct i2c_driver sn3218_driver = {
> > -	.driver = {
> > -		.name	= "leds-sn3218",
> > -		.of_match_table = of_match_ptr(of_sn3218_match),
> > -	},
> > -	.probe	= sn3218_probe,
> > -	.remove	= sn3218_remove,
> > -	.shutdown = sn3218_shutdown,
> > -	.id_table = sn3218_id,
> > -};
> > -
> > -module_i2c_driver(sn3218_driver);
> > -
> > -MODULE_DESCRIPTION("Si-En SN3218 LED Driver");
> > -MODULE_AUTHOR("Stefan Wahren <stefan.wahren@i2se.com>");
> > -MODULE_LICENSE("GPL v2");

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[David Rivshin (Allworx)]

On Fri, 04 Mar 2016 16:38:01 +0100
Jacek Anaszewski <j.anaszewski@samsung.com> wrote:

> On 03/04/2016 03:27 PM, David Rivshin (Allworx) wrote:
> > (Stefan, sorry for the duplicate, I just realized that I originally
> > replied only to you by accident).
> >
> > On Thu, 3 Mar 2016 19:13:03 +0100 (CET)
> > Stefan Wahren <stefan.wahren@i2se.com> wrote:
> >  
> >> Hi David,
> >>
> >> i will test the driver on weekend. Some comments below  
> >
> > Great, thanks very much.
> >  
> >>> "David Rivshin (Allworx)" <drivshin.allworx@gmail.com> hat am 3. März 2016 um
> >>> 04:01 geschrieben:
> >>>
> >>>
> >>> From: David Rivshin <drivshin@allworx.com>
> >>>
> >>> The IS31FL32xx family of LED controllers are I2C devices with multiple
> >>> constant-current channels, each with independent 256-level PWM control.
> >>>
> >>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >>>
> >>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
> >>> (TI am335x) platform.
> >>>
> >>> The programming paradigm of these devices is similar in the following
> >>> ways:
> >>> - All registers are 8 bit
> >>> - All LED control registers are write-only
> >>> - Each LED channel has a PWM register (0-255)
> >>> - PWM register writes are shadowed until an Update register is poked
> >>> - All have a concept of Software Shutdown, which disables output
> >>>
> >>> However, there are some differences in devices:
> >>> - 3236/3235 have a separate Control register for each LED,
> >>> (3218/3216 pack the enable bits into fewer registers)
> >>> - 3236/3235 have a per-channel current divisor setting
> >>> - 3236/3235 have a Global Control register that can turn off all LEDs
> >>> - 3216 is unique in a number of ways
> >>> - OUT9-OUT16 can be configured as GPIOs instead of LED controls
> >>> - LEDs can be programmed with an 8-frame animation, with
> >>> programmable delay between frames
> >>> - LEDs can be modulated by an input audio signal
> >>> - Max output current can be adjusted from 1/4 to 2x globally
> >>> - Has a Configuration register instead of a Shutdown register
> >>>
> >>> This driver currently only supports the base PWM control function
> >>> of these devices. The following features of these devices are not
> >>> implemented, although it should be possible to add them in the future:
> >>> - All devices are capable of going into a lower-power "software
> >>> shutdown" mode.
> >>> - The is31fl3236 and is31fl3235 can reduce the max output current
> >>> per-channel with a divisor of 1, 2, 3, or 4.
> >>> - The is31fl3216 can use some LED channels as GPIOs instead.
> >>> - The is31fl3216 can animate LEDs in hardware.
> >>> - The is31fl3216 can modulate LEDs according to an audio input.
> >>> - The is31fl3216 can reduce/increase max output current globally.
> >>>
> >>> Signed-off-by: David Rivshin <drivshin@allworx.com>
> >>> ---
> >>>
> >>> You may see two instances of this warning:
> >>> "passing argument 1 of 'of_property_read_string' discards 'const'
> >>> qualifier from pointer target type"
> >>> That is a result of of_property_read_string() taking a non-const
> >>> struct device_node pointer parameter. I have separately submitted a
> >>> patch to fix that [1], and a few related functions which had the same
> >>> issue. I'm hoping that will get into linux-next before this does, so
> >>> that the warnings never show up there.
> >>>
> >>> Changes from RFC:
> >>> - Removed max-brightness DT property.
> >>> - Refer to these devices as "LED controllers" in Kconfig.
> >>> - Removed redundant last sentence from Kconfig entry
> >>> - Removed unnecessary debug code.
> >>> - Do not set led_classdev.brightness to 0 explicitly, as it is
> >>> already initialized to 0 by devm_kzalloc().
> >>> - Used of_property_read_string() instead of of_get_property().
> >>> - Fail immediately on DT parsing error in a child node, rather than
> >>> continuing on with the non-faulty ones.
> >>> - Added additional comments for some things that might be non-obvious.
> >>> - Added constants for the location of the SSD bit in the SHUTDOWN
> >>> register, and the 3216's CONFIG register.
> >>> - Added special sw_shutdown_func for the 3216 device, as that bit
> >>> is in a different register, at a different position, and has reverse
> >>> polarity compared to all the other devices.
> >>> - Refactored is31fl32xx_init_regs() to separate out some logic into
> >>> is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
> >>>
> >>> [1] https://lkml.org/lkml/2016/3/2/746
> >>>
> >>> drivers/leds/Kconfig | 8 +
> >>> drivers/leds/Makefile | 1 +
> >>> drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
> >>> 3 files changed, 514 insertions(+)
> >>> create mode 100644 drivers/leds/leds-is31fl32xx.c
> >>>
> >>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> >>> index 1034696..9c63ba4 100644
> >>> --- a/drivers/leds/Kconfig
> >>> +++ b/drivers/leds/Kconfig
> >>> @@ -580,6 +580,14 @@ config LEDS_SN3218
> >>> This driver can also be built as a module. If so the module
> >>> will be called leds-sn3218.
> >>>
> >>> +config LEDS_IS31FL32XX
> >>> + tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
> >>> + depends on LEDS_CLASS && I2C && OF
> >>> + help
> >>> + Say Y here to include support for ISSI IS31FL32XX LED controllers.
> >>> + They are I2C devices with multiple constant-current channels, each
> >>> + with independent 256-level PWM control.  
> >>
> >> Is it worth to mention the module name here?  
> >
> > I noticed that some do and some don't. I don't mind adding it, but it
> > also seemed like it would be obvious, and therefore unnecessary.
> >
> > Jacek, which do you prefer?  
> 
> I agree - it's obvious, we can skip it.
> 
> >>> +
> >>> comment "LED driver for blink(1) USB RGB LED is under Special HID drivers
> >>> (HID_THINGM)"
> >>>
> >>> config LEDS_BLINKM
> >>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
> >>> index 89c9b6f..3fdf313 100644
> >>> --- a/drivers/leds/Makefile
> >>> +++ b/drivers/leds/Makefile
> >>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692) += leds-ktd2692.o
> >>> obj-$(CONFIG_LEDS_POWERNV) += leds-powernv.o
> >>> obj-$(CONFIG_LEDS_SEAD3) += leds-sead3.o
> >>> obj-$(CONFIG_LEDS_SN3218) += leds-sn3218.o
> >>> +obj-$(CONFIG_LEDS_IS31FL32XX) += leds-is31fl32xx.o
> >>>
> >>> # LED SPI Drivers
> >>> obj-$(CONFIG_LEDS_DAC124S085) += leds-dac124s085.o
> >>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
> >>> new file mode 100644
> >>> index 0000000..49818f0
> >>> --- /dev/null
> >>> +++ b/drivers/leds/leds-is31fl32xx.c
> >>> @@ -0,0 +1,505 @@
> >>> +/*
> >>> + * linux/drivers/leds-is31fl32xx.c  
> >>
> >> I think this is unnecessary.  
> >
> > I tend to agree. I think I used leds-pwm.c as a template, and that had
> > such a comment. I assumed it was coding-style and kept it, but now I see
> > that only a minority of led drivers have it. If I do another spin for
> > any reason I'll remove it.
> >  
> >>> + *
> >>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
> >>> + *
> >>> + * Copyright 2015 Allworx Corp.
> >>> + *
> >>> + *
> >>> + * 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.
> >>> + *
> >>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
> >>> + */
> >>> +  
> >>
> >> Shouldn't we include <linux/device.h> here?  
> >
> > Good catch. I was getting that via i2c.h, but since struct device is
> > referenced explicitly in a few places, device.h should probably be
> > included directly.  
> 
> linux/device.h is included from linux/leds.h

This is true, but SubmitChecklist says the following on this topic:
1: If you use a facility then #include the file that defines/declares
   that facility.  Don't depend on other header files pulling in ones
   that you use.
(Oddly CodingStyle is silent on the topic, which is where I would have
thought such a thing would be documented.)

I interpreted that to mean that because 'struct device*' appears in 
this file, it should include <linux/device.h> directly.

> >>> +#include <linux/err.h>
> >>> +#include <linux/i2c.h>
> >>> +#include <linux/kernel.h>
> >>> +#include <linux/leds.h>
> >>> +#include <linux/module.h>
> >>> +#include <linux/of_platform.h>
> >>> +
> >>> +/* Used to indicate a device has no such register */
> >>> +#define IS31FL32XX_REG_NONE 0xFF
> >>> +
> >>> +/* Software Shutdown bit in Shutdown Register */
> >>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE 0
> >>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
> >>> +
> >>> +/* IS31FL3216 has a number of unique registers */
> >>> +#define IS31FL3216_CONFIG_REG 0x00
> >>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
> >>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
> >>> +
> >>> +/* Software Shutdown bit in 3216 Config Register */
> >>> +#define IS31FL3216_CONFIG_SSD_ENABLE BIT(7)
> >>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
> >>> +
> >>> +struct is31fl32xx_priv;
> >>> +struct is31fl32xx_led_data {
> >>> + struct led_classdev cdev;
> >>> + u8 channel; /* 1-based, max priv->cdef->channels */
> >>> + struct is31fl32xx_priv *priv;
> >>> +};
> >>> +
> >>> +struct is31fl32xx_priv {
> >>> + const struct is31fl32xx_chipdef *cdef;
> >>> + struct i2c_client *client;
> >>> + unsigned int num_leds;
> >>> + struct is31fl32xx_led_data leds[0];
> >>> +};
> >>> +
> >>> +/**
> >>> + * struct is31fl32xx_chipdef - chip-specific attributes
> >>> + * @channels : Number of LED channels
> >>> + * @shutdown_reg : address of Shutdown register (optional)
> >>> + * @pwm_update_reg : address of PWM Update register
> >>> + * @global_control_reg : address of Global Control register (optional)
> >>> + * @reset_reg : address of Reset register (optional)
> >>> + * @pwm_register_base : address of first PWM register
> >>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
> >>> + * @led_control_register_base : address of first LED control register
> >>> (optional)
> >>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
> >>> + * @reset_func: : pointer to reset function
> >>> + *
> >>> + * For all optional register addresses, the sentinel value
> >>> %IS31FL32XX_REG_NONE
> >>> + * indicates that this chip has no such register.
> >>> + *
> >>> + * If non-NULL, @reset_func will be called during probing to set all
> >>> + * necessary registers to a known initialization state. This is needed
> >>> + * for chips that do not have a @reset_reg.
> >>> + *
> >>> + * @enable_bits_per_led_control_register must be >=1 if
> >>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
> >>> + */
> >>> +struct is31fl32xx_chipdef {
> >>> + u8 channels;
> >>> + u8 shutdown_reg;
> >>> + u8 pwm_update_reg;
> >>> + u8 global_control_reg;
> >>> + u8 reset_reg;
> >>> + u8 pwm_register_base;
> >>> + bool pwm_registers_reversed;
> >>> + u8 led_control_register_base;
> >>> + u8 enable_bits_per_led_control_register;
> >>> + int (*reset_func)(struct is31fl32xx_priv *priv);
> >>> + int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
> >>> +};
> >>> +
> >>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
> >>> + .channels = 36,
> >>> + .shutdown_reg = 0x00,
> >>> + .pwm_update_reg = 0x25,
> >>> + .global_control_reg = 0x4a,
> >>> + .reset_reg = 0x4f,
> >>> + .pwm_register_base = 0x01,
> >>> + .led_control_register_base = 0x26,
> >>> + .enable_bits_per_led_control_register = 1,
> >>> +};
> >>> +
> >>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
> >>> + .channels = 28,
> >>> + .shutdown_reg = 0x00,
> >>> + .pwm_update_reg = 0x25,
> >>> + .global_control_reg = 0x4a,
> >>> + .reset_reg = 0x4f,
> >>> + .pwm_register_base = 0x05,
> >>> + .led_control_register_base = 0x2a,
> >>> + .enable_bits_per_led_control_register = 1,
> >>> +};
> >>> +
> >>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
> >>> + .channels = 18,
> >>> + .shutdown_reg = 0x00,
> >>> + .pwm_update_reg = 0x16,
> >>> + .global_control_reg = IS31FL32XX_REG_NONE,
> >>> + .reset_reg = 0x17,
> >>> + .pwm_register_base = 0x01,
> >>> + .led_control_register_base = 0x13,
> >>> + .enable_bits_per_led_control_register = 6,
> >>> +};
> >>> +
> >>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
> >>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> >>> + bool enable);
> >>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
> >>> + .channels = 16,
> >>> + .shutdown_reg = IS31FL32XX_REG_NONE,
> >>> + .pwm_update_reg = 0xB0,
> >>> + .global_control_reg = IS31FL32XX_REG_NONE,
> >>> + .reset_reg = IS31FL32XX_REG_NONE,
> >>> + .pwm_register_base = 0x10,
> >>> + .pwm_registers_reversed = true,
> >>> + .led_control_register_base = 0x01,
> >>> + .enable_bits_per_led_control_register = 8,
> >>> + .reset_func = is31fl3216_reset,
> >>> + .sw_shutdown_func = is31fl3216_software_shutdown,
> >>> +};
> >>> +
> >>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
> >>> +{
> >>> + int ret;
> >>> +
> >>> + dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
> >>> +
> >>> + ret = i2c_smbus_write_byte_data(priv->client, reg, val);
> >>> + if (ret) {
> >>> + dev_err(&priv->client->dev,
> >>> + "register write to 0x%02X failed (error %d)",
> >>> + reg, ret);
> >>> + }  
> >>
> >> In case somebody use this driver as heartbeat and writing fails permanently the
> >> log will be flooded.  
> >
> > Unless I'm mistaken that would require the device/bus to fail after
> > successfully probing (probe code itself bails on the first write
> > failure, so there would be no flooding as a result of that). So while
> > not impossible, I imagine it would be unlikely, and I'd hate to remove
> > an error message for such an important condition.
> >
> > I suppose I could use dev_err_ratelimited() to soften any potential
> > flooding, but I second guess that because:
> >   - In led_core.c set_brightness_delayed() has a dev_err() that would come
> >     out on each failed LED update anyways.
> >   - There is precedent in other led drivers of a similar error message.
> >   - Some userspace logging programs will compresses repeated messages anyways.
> >
> > Jacek, what is your preference on this?  
> 
> Let's leave it as is. Permanent I2C bus failure is a critical error and
> flooding the log would only allow to diagnose the problem quicker.
> 
> >>> + return ret;
> >>> +}
> >>> +
> >>> +/*
> >>> + * Custom reset function for IS31FL3216 because it does not have a RESET
> >>> + * register the way that the other IS31FL32xx chips do. We don't bother
> >>> + * writing the GPIO and animation registers, because the registers we
> >>> + * do write ensure those will have no effect.
> >>> + */
> >>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
> >>> +{
> >>> + unsigned int i;
> >>> + int ret;
> >>> +
> >>> + ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
> >>> + IS31FL3216_CONFIG_SSD_ENABLE);
> >>> + if (ret)
> >>> + return ret;
> >>> + for (i = 0; i < priv->cdef->channels; i++) {
> >>> + ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
> >>> + 0x00);
> >>> + if (ret)
> >>> + return ret;
> >>> + }
> >>> + ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
> >>> + if (ret)
> >>> + return ret;
> >>> + ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
> >>> + if (ret)
> >>> + return ret;
> >>> + ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
> >>> + if (ret)
> >>> + return ret;
> >>> +
> >>> + return 0;
> >>> +}
> >>> +
> >>> +/*
> >>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
> >>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
> >>> + * We don't bother doing a read/modify/write on the CONFIG register because
> >>> + * we only ever use a value of '0' for the other fields in that register.
> >>> + */
> >>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
> >>> + bool enable)
> >>> +{
> >>> + u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
> >>> + IS31FL3216_CONFIG_SSD_DISABLE;
> >>> +
> >>> + return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
> >>> +}
> >>> +
> >>> +/*
> >>> + * NOTE: A mutex is not needed in this function because:
> >>> + * - All referenced data is read-only after probe()
> >>> + * - The I2C core has a mutex on to protect the bus
> >>> + * - There are no read/modify/write operations
> >>> + * - Intervening operations between the write of the PWM register
> >>> + * and the Update register are harmless.
> >>> + *
> >>> + * Example:
> >>> + * PWM_REG_1 write 16
> >>> + * UPDATE_REG write 0
> >>> + * PWM_REG_2 write 128
> >>> + * UPDATE_REG write 0
> >>> + * vs:
> >>> + * PWM_REG_1 write 16
> >>> + * PWM_REG_2 write 128
> >>> + * UPDATE_REG write 0
> >>> + * UPDATE_REG write 0
> >>> + * are equivalent. Poking the Update register merely applies all PWM
> >>> + * register writes up to that point.
> >>> + */
> >>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
> >>> + enum led_brightness brightness)
> >>> +{
> >>> + const struct is31fl32xx_led_data *led_data =
> >>> + container_of(led_cdev, struct is31fl32xx_led_data, cdev);
> >>> + const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
> >>> + u8 pwm_register_offset;
> >>> + int ret;
> >>> +
> >>> + dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
> >>> +
> >>> + /* NOTE: led_data->channel is 1-based */
> >>> + if (cdef->pwm_registers_reversed)
> >>> + pwm_register_offset = cdef->channels - led_data->channel;
> >>> + else
> >>> + pwm_register_offset = led_data->channel - 1;
> >>> +
> >>> + ret = is31fl32xx_write(led_data->priv,
> >>> + cdef->pwm_register_base + pwm_register_offset,
> >>> + brightness);
> >>> + if (ret)
> >>> + return ret;
> >>> +
> >>> + return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
> >>> +}
> >>> +
> >>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
> >>> +{
> >>> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>> + int ret;
> >>> +
> >>> + if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
> >>> + ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
> >>> + if (ret)
> >>> + return ret;
> >>> + }
> >>> +
> >>> + if (cdef->reset_func)
> >>> + return cdef->reset_func(priv);
> >>> +
> >>> + return 0;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
> >>> + bool enable)
> >>> +{
> >>> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>> + int ret;
> >>> +
> >>> + if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
> >>> + u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
> >>> + IS31FL32XX_SHUTDOWN_SSD_DISABLE;
> >>> + ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
> >>> + if (ret)
> >>> + return ret;
> >>> + }
> >>> +
> >>> + if (cdef->sw_shutdown_func)
> >>> + return cdef->sw_shutdown_func(priv, enable);
> >>> +
> >>> + return 0;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
> >>> +{
> >>> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
> >>> + int ret;
> >>> +
> >>> + ret = is31fl32xx_reset_regs(priv);
> >>> + if (ret)
> >>> + return ret;
> >>> +
> >>> + /*
> >>> + * Set enable bit for all channels.
> >>> + * We will control state with PWM registers alone.
> >>> + */
> >>> + if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
> >>> + u8 value =
> >>> + GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
> >>> + u8 num_regs = cdef->channels /
> >>> + cdef->enable_bits_per_led_control_register;
> >>> + int i;
> >>> +
> >>> + for (i = 0; i < num_regs; i++) {
> >>> + ret = is31fl32xx_write(priv,
> >>> + cdef->led_control_register_base+i,
> >>> + value);
> >>> + if (ret)
> >>> + return ret;
> >>> + }
> >>> + }
> >>> +
> >>> + ret = is31fl32xx_software_shutdown(priv, false);
> >>> + if (ret)
> >>> + return ret;
> >>> +
> >>> + if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
> >>> + ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
> >>> + if (ret)
> >>> + return ret;
> >>> + }
> >>> +
> >>> + return 0;
> >>> +}
> >>> +
> >>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
> >>> +{
> >>> + return sizeof(struct is31fl32xx_priv) +
> >>> + (sizeof(struct is31fl32xx_led_data) * num_leds);
> >>> +}
> >>> +
> >>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
> >>> + const struct device_node *child,
> >>> + struct is31fl32xx_led_data *led_data)
> >>> +{
> >>> + struct led_classdev *cdev = &led_data->cdev;
> >>> + int ret = 0;
> >>> + u32 reg;
> >>> +
> >>> + if (of_property_read_string(child, "label", &cdev->name))
> >>> + cdev->name = child->name;
> >>> +
> >>> + ret = of_property_read_u32(child, "reg", &reg);
> >>> + if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
> >>> + dev_err(dev,
> >>> + "Child node %s does not have a valid reg property\n",
> >>> + child->full_name);
> >>> + return -EINVAL;
> >>> + }
> >>> + led_data->channel = reg;
> >>> +
> >>> + of_property_read_string(child, "linux,default-trigger",
> >>> + &cdev->default_trigger);
> >>> +
> >>> + cdev->brightness_set_blocking = is31fl32xx_brightness_set;
> >>> +
> >>> + return 0;
> >>> +}
> >>> +
> >>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
> >>> + struct is31fl32xx_priv *priv,
> >>> + u8 channel)
> >>> +{
> >>> + size_t i;
> >>> +
> >>> + for (i = 0; i < priv->num_leds; i++) {
> >>> + if (priv->leds[i].channel == channel)
> >>> + return &priv->leds[i];
> >>> + }
> >>> +
> >>> + return NULL;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_parse_dt(struct device *dev,
> >>> + struct is31fl32xx_priv *priv)
> >>> +{
> >>> + struct device_node *child;
> >>> + int ret = 0;
> >>> +
> >>> + for_each_child_of_node(dev->of_node, child) {
> >>> + struct is31fl32xx_led_data *led_data =
> >>> + &priv->leds[priv->num_leds];  
> >>
> >> Maybe i missed something, but is it really protected against out of index
> >> access?  
> >
> > The array is allocated with size equal to the number of child nodes,
> > and num_leds is incremented once for each child node parsed. So in
> > order for the index to be out of bounds, the number of child nodes
> > would need to increase during the probe. I assumed that the DT is
> > static during probing, but if that's not the case then you're right
> > that this is a potential problem. Also, this equivalent logic is
> > used in leds-pwm, leds-gpio, and leds-ns2, so that gives me
> > confidence that its safe.
> > Unless DT overlays change that assumption?  
> 
> DT overlays would matter here if child DT nodes could be dynamically
> removed, i.e. if it was possible by design to dynamically unplug LEDs
> from the current outputs during LED controller operation, which is not
> the case for this device (and any other LED controller I am aware of).
> 
> >>> + const struct is31fl32xx_led_data *other_led_data;
> >>> +
> >>> + led_data->priv = priv;
> >>> +
> >>> + ret = is31fl32xx_parse_child_dt(dev, child, led_data);
> >>> + if (ret)
> >>> + goto err;
> >>> +
> >>> + /* Detect if channel is already in use by another child */
> >>> + other_led_data = is31fl32xx_find_led_data(priv,
> >>> + led_data->channel);
> >>> + if (other_led_data) {
> >>> + dev_err(dev,
> >>> + "%s and %s both attempting to use channel %d\n",
> >>> + led_data->cdev.name,
> >>> + other_led_data->cdev.name,
> >>> + led_data->channel);
> >>> + goto err;
> >>> + }
> >>> +
> >>> + ret = devm_led_classdev_register(dev, &led_data->cdev);
> >>> + if (ret) {
> >>> + dev_err(dev, "failed to register PWM led for %s: %d\n",
> >>> + led_data->cdev.name, ret);
> >>> + goto err;
> >>> + }
> >>> +
> >>> + priv->num_leds++;
> >>> + }
> >>> +
> >>> + return 0;
> >>> +
> >>> +err:
> >>> + of_node_put(child);
> >>> + return ret;
> >>> +}
> >>> +
> >>> +static const struct of_device_id of_is31fl31xx_match[] = {
> >>> + { .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
> >>> + { .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
> >>> + { .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
> >>> + { .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
> >>> + {},
> >>> +};
> >>> +
> >>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
> >>> +
> >>> +static int is31fl32xx_probe(struct i2c_client *client,
> >>> + const struct i2c_device_id *id)
> >>> +{
> >>> + const struct is31fl32xx_chipdef *cdef;
> >>> + const struct of_device_id *of_dev_id;
> >>> + struct device *dev = &client->dev;
> >>> + struct is31fl32xx_priv *priv;
> >>> + int count;
> >>> + int ret = 0;
> >>> +
> >>> + of_dev_id = of_match_device(of_is31fl31xx_match, dev);
> >>> + if (!of_dev_id)
> >>> + return -EINVAL;
> >>> +
> >>> + cdef = of_dev_id->data;
> >>> +
> >>> + count = of_get_child_count(dev->of_node);
> >>> + if (!count)
> >>> + return -EINVAL;
> >>> +
> >>> + priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
> >>> + GFP_KERNEL);
> >>> + if (!priv)
> >>> + return -ENOMEM;
> >>> +
> >>> + priv->client = client;
> >>> + priv->cdef = cdef;
> >>> + i2c_set_clientdata(client, priv);
> >>> +
> >>> + ret = is31fl32xx_init_regs(priv);
> >>> + if (ret)
> >>> + return ret;
> >>> +
> >>> + ret = is31fl32xx_parse_dt(dev, priv);
> >>> + if (ret)
> >>> + return ret;
> >>> +
> >>> + return 0;
> >>> +}
> >>> +
> >>> +static int is31fl32xx_remove(struct i2c_client *client)
> >>> +{
> >>> + struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
> >>> +
> >>> + return is31fl32xx_reset_regs(priv);
> >>> +}
> >>> +
> >>> +/*
> >>> + * i2c-core requires that id_table be non-NULL, even though
> >>> + * it is not used for DeviceTree based instantiation.
> >>> + */
> >>> +static const struct i2c_device_id is31fl31xx_id[] = {
> >>> + {},
> >>> +};
> >>> +
> >>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
> >>> +
> >>> +static struct i2c_driver is31fl32xx_driver = {
> >>> + .driver = {
> >>> + .name = "is31fl32xx",
> >>> + .of_match_table = of_is31fl31xx_match,
> >>> + },
> >>> + .probe = is31fl32xx_probe,
> >>> + .remove = is31fl32xx_remove,  
> >>
> >> Sorry, what was the reason to skip shutdown?  
> >
> > If I understood Jacek's last email on the topic [1] correctly, he's now
> > of the opinion that the decision to turn LEDs off on reboot should be
> > left to userspace, rather than done by the driver. For these devices,
> > the only thing a shutdown callback would do is turn off the LEDs (through
> > any of multiple methods). So, if we want to leave the state as-is on
> > reboot there's no need for a shutdown callback.
> >
> > [1] http://www.spinics.net/lists/linux-leds/msg05644.html
> >  
> >>> + .id_table = is31fl31xx_id,
> >>> +};
> >>> +
> >>> +module_i2c_driver(is31fl32xx_driver);
> >>> +
> >>> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
> >>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
> >>> +MODULE_LICENSE("GPL v2");
> >>> --
> >>> 2.5.0
> >>>  

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[Jacek Anaszewski]

On 03/05/2016 06:29 AM, David Rivshin (Allworx) wrote:
> On Fri, 4 Mar 2016 22:39:06 +0100
> Jacek Anaszewski <jacek.anaszewski@gmail.com> wrote:
>
>> On 03/04/2016 08:02 PM, David Rivshin (Allworx) wrote:
>>> On Fri, 04 Mar 2016 17:01:52 +0100
>>> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>>>
>>>> On 03/04/2016 04:05 PM, David Rivshin (Allworx) wrote:
>>>>> On Fri, 04 Mar 2016 08:54:02 +0100
>>>>> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>>>>>
>>>>>> On 03/04/2016 01:45 AM, David Rivshin (Allworx) wrote:
>>>>>>> On Thu, 03 Mar 2016 15:51:32 +0100
>>>>>>> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>>>>>>>
>>>>>>>> Hi David,
>>>>>>>>
>>>>>>>> Thanks for the update. Two remarks in the code.
>>>>>>>>
>>>>>>>> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:
>>>>>>>>> From: David Rivshin <drivshin@allworx.com>
>>>>>>>>>
>>>>>>>>> The IS31FL32xx family of LED controllers are I2C devices with multiple
>>>>>>>>> constant-current channels, each with independent 256-level PWM control.
>>>>>>>>>
>>>>>>>>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>>>>>>>>
>>>>>>>>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
>>>>>>>>> (TI am335x) platform.
>>>>>>>>>
>>>>>>>>> The programming paradigm of these devices is similar in the following
>>>>>>>>> ways:
>>>>>>>>>       - All registers are 8 bit
>>>>>>>>>       - All LED control registers are write-only
>>>>>>>>>       - Each LED channel has a PWM register (0-255)
>>>>>>>>>       - PWM register writes are shadowed until an Update register is poked
>>>>>>>>>       - All have a concept of Software Shutdown, which disables output
>>>>>>>>>
>>>>>>>>> However, there are some differences in devices:
>>>>>>>>>       - 3236/3235 have a separate Control register for each LED,
>>>>>>>>>         (3218/3216 pack the enable bits into fewer registers)
>>>>>>>>>       - 3236/3235 have a per-channel current divisor setting
>>>>>>>>>       - 3236/3235 have a Global Control register that can turn off all LEDs
>>>>>>>>>       - 3216 is unique in a number of ways
>>>>>>>>>          - OUT9-OUT16 can be configured as GPIOs instead of LED controls
>>>>>>>>>          - LEDs can be programmed with an 8-frame animation, with
>>>>>>>>>            programmable delay between frames
>>>>>>>>>          - LEDs can be modulated by an input audio signal
>>>>>>>>>          - Max output current can be adjusted from 1/4 to 2x globally
>>>>>>>>>          - Has a Configuration register instead of a Shutdown register
>>>>>>>>>
>>>>>>>>> This driver currently only supports the base PWM control function
>>>>>>>>> of these devices. The following features of these devices are not
>>>>>>>>> implemented, although it should be possible to add them in the future:
>>>>>>>>>       - All devices are capable of going into a lower-power "software
>>>>>>>>>         shutdown" mode.
>>>>>>>>>       - The is31fl3236 and is31fl3235 can reduce the max output current
>>>>>>>>>         per-channel with a divisor of 1, 2, 3, or 4.
>>>>>>>>>       - The is31fl3216 can use some LED channels as GPIOs instead.
>>>>>>>>>       - The is31fl3216 can animate LEDs in hardware.
>>>>>>>>>       - The is31fl3216 can modulate LEDs according to an audio input.
>>>>>>>>>       - The is31fl3216 can reduce/increase max output current globally.
>>>>>>>>>
>>>>>>>>> Signed-off-by: David Rivshin <drivshin@allworx.com>
>>>>>>>>> ---
>>>>>>>>>
>>>>>>>>> You may see two instances of this warning:
>>>>>>>>>        "passing argument 1 of 'of_property_read_string' discards 'const'
>>>>>>>>>         qualifier from pointer target type"
>>>>>>>>> That is a result of of_property_read_string() taking a non-const
>>>>>>>>> struct device_node pointer parameter. I have separately submitted a
>>>>>>>>> patch to fix that [1], and a few related functions which had the same
>>>>>>>>> issue. I'm hoping that will get into linux-next before this does, so
>>>>>>>>> that the warnings never show up there.
>>>>>>>>
>>>>>>>> Please adjust the patch so that it compiles without warnings on
>>>>>>>> current linux-next. Your patch for DT API hasn't been reviewed yet
>>>>>>>> AFICS, and I can imagine that there will be some resistance against.
>>>>>>>
>>>>>>> Since the DT API patch was just accepted by Rob [1], would it be OK
>>>>>>> to wait for the results of Stefan's testing (and any other reviews)
>>>>>>> before making a decision on this? From Stefan's note, it won't be
>>>>>>> until this weekend that he will have a chance to test, and I'm
>>>>>>> guessing the DT API patch will make its way through Rob's tree to
>>>>>>> linux-next by then.
>>>>>>
>>>>>> OK.
>>>>>>
>>>>>>> FYI, the warning workaround would be to make the second parameter to
>>>>>>> is31fl32xx_parse_child_dt() non-const.
>>>>>>>
>>>>>>> [1] https://lkml.org/lkml/2016/3/3/924
>>>>>>>
>>>>>>>>> Changes from RFC:
>>>>>>>>>       - Removed max-brightness DT property.
>>>>>>>>>       - Refer to these devices as "LED controllers" in Kconfig.
>>>>>>>>>       - Removed redundant last sentence from Kconfig entry
>>>>>>>>>       - Removed unnecessary debug code.
>>>>>>>>>       - Do not set led_classdev.brightness to 0 explicitly, as it is
>>>>>>>>>         already initialized to 0 by devm_kzalloc().
>>>>>>>>>       - Used of_property_read_string() instead of of_get_property().
>>>>>>>>>       - Fail immediately on DT parsing error in a child node, rather than
>>>>>>>>>         continuing on with the non-faulty ones.
>>>>>>>>>       - Added additional comments for some things that might be non-obvious.
>>>>>>>>>       - Added constants for the location of the SSD bit in the SHUTDOWN
>>>>>>>>>         register, and the 3216's CONFIG register.
>>>>>>>>>       - Added special sw_shutdown_func for the 3216 device, as that bit
>>>>>>>>>         is in a different register, at a different position, and has reverse
>>>>>>>>>         polarity compared to all the other devices.
>>>>>>>>>       - Refactored is31fl32xx_init_regs() to separate out some logic into
>>>>>>>>>         is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
>>>>>>>>>
>>>>>>>>> [1] https://lkml.org/lkml/2016/3/2/746
>>>>>>>>>
>>>>>>>>>       drivers/leds/Kconfig           |   8 +
>>>>>>>>>       drivers/leds/Makefile          |   1 +
>>>>>>>>>       drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
>>>>>>>>>       3 files changed, 514 insertions(+)
>>>>>>>>>       create mode 100644 drivers/leds/leds-is31fl32xx.c
>>>>>>>>>
>>>>>>>>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
>>>>>>>>> index 1034696..9c63ba4 100644
>>>>>>>>> --- a/drivers/leds/Kconfig
>>>>>>>>> +++ b/drivers/leds/Kconfig
>>>>>>>>> @@ -580,6 +580,14 @@ config LEDS_SN3218
>>>>>>>>>       	  This driver can also be built as a module. If so the module
>>>>>>>>>       	  will be called leds-sn3218.
>>>>>>>>>
>>>>>>>>> +config LEDS_IS31FL32XX
>>>>>>>>> +	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
>>>>>>>>> +	depends on LEDS_CLASS && I2C && OF
>>>>>>>>> +	help
>>>>>>>>> +	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
>>>>>>>>> +	  They are I2C devices with multiple constant-current channels, each
>>>>>>>>> +	  with independent 256-level PWM control.
>>>>>>>>> +
>>>>>>>>>       comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
>>>>>>>>>
>>>>>>>>>       config LEDS_BLINKM
>>>>>>>>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
>>>>>>>>> index 89c9b6f..3fdf313 100644
>>>>>>>>> --- a/drivers/leds/Makefile
>>>>>>>>> +++ b/drivers/leds/Makefile
>>>>>>>>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
>>>>>>>>>       obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
>>>>>>>>>       obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
>>>>>>>>>       obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
>>>>>>>>> +obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
>>>>>>>>>
>>>>>>>>>       # LED SPI Drivers
>>>>>>>>>       obj-$(CONFIG_LEDS_DAC124S085)		+= leds-dac124s085.o
>>>>>>>>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
>>>>>>>>> new file mode 100644
>>>>>>>>> index 0000000..49818f0
>>>>>>>>> --- /dev/null
>>>>>>>>> +++ b/drivers/leds/leds-is31fl32xx.c
>>>>>>>>> @@ -0,0 +1,505 @@
>>>>>>>>> +/*
>>>>>>>>> + * linux/drivers/leds-is31fl32xx.c
>>>>>>>>> + *
>>>>>>>>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
>>>>>>>>> + *
>>>>>>>>> + * Copyright 2015 Allworx Corp.
>>>>>>>>> + *
>>>>>>>>> + *
>>>>>>>>> + * 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.
>>>>>>>>> + *
>>>>>>>>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>>>>>>>> + */
>>>>>>>>> +
>>>>>>>>> +#include <linux/err.h>
>>>>>>>>> +#include <linux/i2c.h>
>>>>>>>>> +#include <linux/kernel.h>
>>>>>>>>> +#include <linux/leds.h>
>>>>>>>>> +#include <linux/module.h>
>>>>>>>>> +#include <linux/of_platform.h>
>>>>>>>>> +
>>>>>>>>> +/* Used to indicate a device has no such register */
>>>>>>>>> +#define IS31FL32XX_REG_NONE 0xFF
>>>>>>>>> +
>>>>>>>>> +/* Software Shutdown bit in Shutdown Register */
>>>>>>>>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
>>>>>>>>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
>>>>>>>>> +
>>>>>>>>> +/* IS31FL3216 has a number of unique registers */
>>>>>>>>> +#define IS31FL3216_CONFIG_REG 0x00
>>>>>>>>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
>>>>>>>>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
>>>>>>>>> +
>>>>>>>>> +/* Software Shutdown bit in 3216 Config Register */
>>>>>>>>> +#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
>>>>>>>>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
>>>>>>>>> +
>>>>>>>>> +struct is31fl32xx_priv;
>>>>>>>>> +struct is31fl32xx_led_data {
>>>>>>>>> +	struct led_classdev cdev;
>>>>>>>>> +	u8 channel; /* 1-based, max priv->cdef->channels */
>>>>>>>>> +	struct is31fl32xx_priv *priv;
>>>>>>>>> +};
>>>>>>>>> +
>>>>>>>>> +struct is31fl32xx_priv {
>>>>>>>>> +	const struct is31fl32xx_chipdef *cdef;
>>>>>>>>> +	struct i2c_client *client;
>>>>>>>>> +	unsigned int num_leds;
>>>>>>>>> +	struct is31fl32xx_led_data leds[0];
>>>>>>>>> +};
>>>>>>>>> +
>>>>>>>>> +/**
>>>>>>>>> + * struct is31fl32xx_chipdef - chip-specific attributes
>>>>>>>>> + * @channels            : Number of LED channels
>>>>>>>>> + * @shutdown_reg        : address of Shutdown register (optional)
>>>>>>>>> + * @pwm_update_reg      : address of PWM Update register
>>>>>>>>> + * @global_control_reg  : address of Global Control register (optional)
>>>>>>>>> + * @reset_reg           : address of Reset register (optional)
>>>>>>>>> + * @pwm_register_base   : address of first PWM register
>>>>>>>>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
>>>>>>>>> + * @led_control_register_base : address of first LED control register (optional)
>>>>>>>>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
>>>>>>>>> + * @reset_func:         : pointer to reset function
>>>>>>>>> + *
>>>>>>>>> + * For all optional register addresses, the sentinel value %IS31FL32XX_REG_NONE
>>>>>>>>> + * indicates that this chip has no such register.
>>>>>>>>> + *
>>>>>>>>> + * If non-NULL, @reset_func will be called during probing to set all
>>>>>>>>> + * necessary registers to a known initialization state. This is needed
>>>>>>>>> + * for chips that do not have a @reset_reg.
>>>>>>>>> + *
>>>>>>>>> + * @enable_bits_per_led_control_register must be >=1 if
>>>>>>>>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
>>>>>>>>> + */
>>>>>>>>> +struct is31fl32xx_chipdef {
>>>>>>>>> +	u8	channels;
>>>>>>>>> +	u8	shutdown_reg;
>>>>>>>>> +	u8	pwm_update_reg;
>>>>>>>>> +	u8	global_control_reg;
>>>>>>>>> +	u8	reset_reg;
>>>>>>>>> +	u8	pwm_register_base;
>>>>>>>>> +	bool	pwm_registers_reversed;
>>>>>>>>> +	u8	led_control_register_base;
>>>>>>>>> +	u8	enable_bits_per_led_control_register;
>>>>>>>>> +	int (*reset_func)(struct is31fl32xx_priv *priv);
>>>>>>>>> +	int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
>>>>>>>>> +};
>>>>>>>>> +
>>>>>>>>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
>>>>>>>>> +	.channels				= 36,
>>>>>>>>> +	.shutdown_reg				= 0x00,
>>>>>>>>> +	.pwm_update_reg				= 0x25,
>>>>>>>>> +	.global_control_reg			= 0x4a,
>>>>>>>>> +	.reset_reg				= 0x4f,
>>>>>>>>> +	.pwm_register_base			= 0x01,
>>>>>>>>> +	.led_control_register_base		= 0x26,
>>>>>>>>> +	.enable_bits_per_led_control_register	= 1,
>>>>>>>>> +};
>>>>>>>>> +
>>>>>>>>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
>>>>>>>>> +	.channels				= 28,
>>>>>>>>> +	.shutdown_reg				= 0x00,
>>>>>>>>> +	.pwm_update_reg				= 0x25,
>>>>>>>>> +	.global_control_reg			= 0x4a,
>>>>>>>>> +	.reset_reg				= 0x4f,
>>>>>>>>> +	.pwm_register_base			= 0x05,
>>>>>>>>> +	.led_control_register_base		= 0x2a,
>>>>>>>>> +	.enable_bits_per_led_control_register	= 1,
>>>>>>>>> +};
>>>>>>>>> +
>>>>>>>>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
>>>>>>>>> +	.channels				= 18,
>>>>>>>>> +	.shutdown_reg				= 0x00,
>>>>>>>>> +	.pwm_update_reg				= 0x16,
>>>>>>>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
>>>>>>>>> +	.reset_reg				= 0x17,
>>>>>>>>> +	.pwm_register_base			= 0x01,
>>>>>>>>> +	.led_control_register_base		= 0x13,
>>>>>>>>> +	.enable_bits_per_led_control_register	= 6,
>>>>>>>>> +};
>>>>>>>>> +
>>>>>>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
>>>>>>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>>>>>>>> +					bool enable);
>>>>>>>>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
>>>>>>>>> +	.channels				= 16,
>>>>>>>>> +	.shutdown_reg				= IS31FL32XX_REG_NONE,
>>>>>>>>> +	.pwm_update_reg				= 0xB0,
>>>>>>>>> +	.global_control_reg			= IS31FL32XX_REG_NONE,
>>>>>>>>> +	.reset_reg				= IS31FL32XX_REG_NONE,
>>>>>>>>> +	.pwm_register_base			= 0x10,
>>>>>>>>> +	.pwm_registers_reversed			= true,
>>>>>>>>> +	.led_control_register_base		= 0x01,
>>>>>>>>> +	.enable_bits_per_led_control_register	= 8,
>>>>>>>>> +	.reset_func				= is31fl3216_reset,
>>>>>>>>> +	.sw_shutdown_func			= is31fl3216_software_shutdown,
>>>>>>>>> +};
>>>>>>>>> +
>>>>>>>>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
>>>>>>>>> +{
>>>>>>>>> +	int ret;
>>>>>>>>> +
>>>>>>>>> +	dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
>>>>>>>>> +
>>>>>>>>> +	ret =  i2c_smbus_write_byte_data(priv->client, reg, val);
>>>>>>>>> +	if (ret) {
>>>>>>>>> +		dev_err(&priv->client->dev,
>>>>>>>>> +			"register write to 0x%02X failed (error %d)",
>>>>>>>>> +			reg, ret);
>>>>>>>>> +	}
>>>>>>>>> +	return ret;
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +/*
>>>>>>>>> + * Custom reset function for IS31FL3216 because it does not have a RESET
>>>>>>>>> + * register the way that the other IS31FL32xx chips do. We don't bother
>>>>>>>>> + * writing the GPIO and animation registers, because the registers we
>>>>>>>>> + * do write ensure those will have no effect.
>>>>>>>>> + */
>>>>>>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
>>>>>>>>> +{
>>>>>>>>> +	unsigned int i;
>>>>>>>>> +	int ret;
>>>>>>>>> +
>>>>>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
>>>>>>>>> +			       IS31FL3216_CONFIG_SSD_ENABLE);
>>>>>>>>> +	if (ret)
>>>>>>>>> +		return ret;
>>>>>>>>> +	for (i = 0; i < priv->cdef->channels; i++) {
>>>>>>>>> +		ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
>>>>>>>>> +				       0x00);
>>>>>>>>> +		if (ret)
>>>>>>>>> +			return ret;
>>>>>>>>> +	}
>>>>>>>>> +	ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
>>>>>>>>> +	if (ret)
>>>>>>>>> +		return ret;
>>>>>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
>>>>>>>>> +	if (ret)
>>>>>>>>> +		return ret;
>>>>>>>>> +	ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
>>>>>>>>> +	if (ret)
>>>>>>>>> +		return ret;
>>>>>>>>> +
>>>>>>>>> +	return 0;
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +/*
>>>>>>>>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
>>>>>>>>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
>>>>>>>>> + * We don't bother doing a read/modify/write on the CONFIG register because
>>>>>>>>> + * we only ever use a value of '0' for the other fields in that register.
>>>>>>>>> + */
>>>>>>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>>>>>>>> +					bool enable)
>>>>>>>>> +{
>>>>>>>>> +	u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
>>>>>>>>> +			    IS31FL3216_CONFIG_SSD_DISABLE;
>>>>>>>>> +
>>>>>>>>> +	return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +/*
>>>>>>>>> + * NOTE: A mutex is not needed in this function because:
>>>>>>>>> + * - All referenced data is read-only after probe()
>>>>>>>>> + * - The I2C core has a mutex on to protect the bus
>>>>>>>>> + * - There are no read/modify/write operations
>>>>>>>>> + * - Intervening operations between the write of the PWM register
>>>>>>>>> + *   and the Update register are harmless.
>>>>>>>>> + *
>>>>>>>>> + * Example:
>>>>>>>>> + *	PWM_REG_1 write 16
>>>>>>>>> + *	UPDATE_REG write 0
>>>>>>>>> + *	PWM_REG_2 write 128
>>>>>>>>> + *	UPDATE_REG write 0
>>>>>>>>> + *   vs:
>>>>>>>>> + *	PWM_REG_1 write 16
>>>>>>>>> + *	PWM_REG_2 write 128
>>>>>>>>> + *	UPDATE_REG write 0
>>>>>>>>> + *	UPDATE_REG write 0
>>>>>>>>> + * are equivalent. Poking the Update register merely applies all PWM
>>>>>>>>> + * register writes up to that point.
>>>>>>>>> + */
>>>>>>>>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
>>>>>>>>> +				     enum led_brightness brightness)
>>>>>>>>> +{
>>>>>>>>> +	const struct is31fl32xx_led_data *led_data =
>>>>>>>>> +		container_of(led_cdev, struct is31fl32xx_led_data, cdev);
>>>>>>>>> +	const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
>>>>>>>>> +	u8 pwm_register_offset;
>>>>>>>>> +	int ret;
>>>>>>>>> +
>>>>>>>>> +	dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
>>>>>>>>> +
>>>>>>>>> +	/* NOTE: led_data->channel is 1-based */
>>>>>>>>> +	if (cdef->pwm_registers_reversed)
>>>>>>>>> +		pwm_register_offset = cdef->channels - led_data->channel;
>>>>>>>>> +	else
>>>>>>>>> +		pwm_register_offset = led_data->channel - 1;
>>>>>>>>> +
>>>>>>>>> +	ret = is31fl32xx_write(led_data->priv,
>>>>>>>>> +			       cdef->pwm_register_base + pwm_register_offset,
>>>>>>>>> +			       brightness);
>>>>>>>>> +	if (ret)
>>>>>>>>> +		return ret;
>>>>>>>>> +
>>>>>>>>> +	return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
>>>>>>>>> +{
>>>>>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>>>>>> +	int ret;
>>>>>>>>> +
>>>>>>>>> +	if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
>>>>>>>>> +		ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
>>>>>>>>> +		if (ret)
>>>>>>>>> +			return ret;
>>>>>>>>> +	}
>>>>>>>>> +
>>>>>>>>> +	if (cdef->reset_func)
>>>>>>>>> +		return cdef->reset_func(priv);
>>>>>>>>> +
>>>>>>>>> +	return 0;
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
>>>>>>>>> +					bool enable)
>>>>>>>>> +{
>>>>>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>>>>>> +	int ret;
>>>>>>>>> +
>>>>>>>>> +	if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
>>>>>>>>> +		u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
>>>>>>>>> +				    IS31FL32XX_SHUTDOWN_SSD_DISABLE;
>>>>>>>>> +		ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
>>>>>>>>> +		if (ret)
>>>>>>>>> +			return ret;
>>>>>>>>> +	}
>>>>>>>>> +
>>>>>>>>> +	if (cdef->sw_shutdown_func)
>>>>>>>>> +		return cdef->sw_shutdown_func(priv, enable);
>>>>>>>>
>>>>>>>> You seem to call sw_shutdown_func only here, so why should we have
>>>>>>>> enable parameter in this op?
>>>>>>>
>>>>>>> I'm not sure if I understand the question, but I will try to answer.
>>>>>>>
>>>>>>> 'enable' is passed through is31fl32xx_software_shutdown to
>>>>>>> cdef->sw_shutdown_func, so it can be either true or false at that
>>>>>>> point. The purpose of sw_shutdown_func is to add any special behavior
>>>>>>> when enabling/disabling software-shutdown mode, which is needed for
>>>>>>> the 3216 because its SSD bit is in a different position and with
>>>>>>> opposite polarity.
>>>>>>>
>>>>>>> Is it that 'enable' in that line of code makes it look like it's being
>>>>>>> called with an hardcoded value rather than a variable? If so, perhaps a
>>>>>>> different parameter name would make it more obvious? Or a kerneldoc
>>>>>>> comment for the function to describe the parameter?
>>>>>>>
>>>>>>> Or have I totally missed the point of the question?
>>>>>>
>>>>>> Actually I should have placed this question next to
>>>>>> the call to s31fl32xx_software_shutdown() in is31fl32xx_init_regs(),
>>>>>> which is passed "false" in the second argument, and there is no
>>>>>> other call to s31fl32xx_software_shutdown() in the driver.
>>>>>>
>>>>>> Having the argument makes people wondering that there is some
>>>>>> use case in the driver, where "true" is passed, but it seems not
>>>>>> to be the case.
>>>>>
>>>>> Thanks for the clarification.
>>>>>
>>>>> Yes, there is currently no explicit call to enable software-shutdown
>>>>> mode. Since the reset state of these devices is to have software-shutdown
>>>>> enabled, the only explicit operation on it that's currently needed is
>>>>> to disable it.
>>>>>
>>>>> Reasons I can think to have the 'enable' parameter anyways include:
>>>>> - It seems logical to me that an API to manipulate the software-shutdown
>>>>>      state should allow both enabling and disabling.
>>>>> - Having a parameter for that seemed the most obvious way to go, and it
>>>>>      was trivial to implement.
>>>>> - Alternatively having separate "enable" and "disable" functions would
>>>>>      duplicate most of the logic, vs the parameter being just a single
>>>>>      conditional. And that would also imply two function pointers in the
>>>>>      chipdefs, which I'd prefer to minimize.
>>>>> - If anyone wanted to implement suspend/resume in the future, they would
>>>>>      most likely do it by enabling/disabling software-shutdown. Supporting
>>>>>      both enable/disable from the start should make that trivial to do.
>>>>
>>>> Suspend/resume callbacks are already implemented in led-class.c and
>>>> related ops indirectly call brightness_set. If you want to support
>>>> suspend/resume you have to set LED_CORE_SUSPENDRESUME flag.
>>>
>>> If I understand correctly, all LED_CORE_SUSPENDRESUME will do is cause
>>> the led core to set the brightness to 0 on suspend (and reverse that
>>> on resume). I see some drivers use this flag and other do not.
>>> This brings up the question in my mind: how would a driver decide
>>> whether it is  appropriate for an LED to go dark on suspend? Is that
>>> just the drivers that know the logical purpose of the LEDs they control?
>>
>> There is a room for improvement here. Possibly a new LED class sysfs
>> attribute could be of help in determining that.
>>
>>>> Setting brightness to 0 is equivalent to turning the LED controller
>>>> in a shutdown mode, provided that all sub-LEDs are off.
>>>
>>> This is not strictly true for these devices. If someone cared very much
>>> about power usage when suspended they may want to put the LED controller
>>> into what it calls "shutdown mode" via the SSD bit. I didn't bother mainly
>>> because I have no need, and also because I wasn't sure how to even trigger
>>> a suspend in order to test an implementation.
>>
>> I meant that LED class driver should put the device in a software
>> shutdown mode after last sub-LED is turned off.
>>
>>> FYI, I just measured it the effect of software-shutdown even with all LEDs
>>> already off. The difference in current is about 5mA, measured at the 5V
>>> supply to a 3216 eval board. Not huge, but someone might care.
>>
>> This can be vital difference for some use cases. You could count the
>> number of currently active sub-LEDs and put the controller in a software
>> shutdown mode in case the value is 0.
>
> I had thought about that, but I had some concerns that dissuaded me from
> perusing it:
> - Would need some way to know whether a brightness_set has changed an LED
>    from on->off or off->on in order to update that counter. leds-core
>    modifies led_cdev->brightness before calling brightness_set, so that
>    information would need to be somehow maintained in the driver.

We could think of adding some generic mechanism to the LED core in the
future.

> - It requires an additional mutex to protect whatever data is involved in
>    the previous bullet. That mutex would need to be taken on every brightness
>    update.
> - If a single LED is blinking rapidly (e.g. a cpu trigger), it would thrash
>    software shutdown mode. That in turn adds even more CPU usage (due to
>    register writes and mutexes), potentially making power usage worse in
>    some cases.
> - Time it might take to implement and review (mostly depending on how the
>    first bullet is handled), especially since the 4.6 merge window is fast
>    approaching.

Thanks for this analysis. The issue turns out to be non-trivial.

> That said, I could propose the following simple-to-implement design:
> - counter, and mutex to protect it, added to 'priv'
> - previous_brightness field added to 'led_data'
> - brightness_set compares new brightness vs previous
>     - if on/off state has changed:
>       - takes mutex
>       - updates counter
>       - sets software-shutdown mode to (counter==0)
>       - releases mutex
>
> If you think that's worth doing, I can try to get it in along with the
> other changes for v2.

This can be added as an incremental patch. As you noticed, the merge
window is approaching and it would be good to have this driver in
linux-next for some time before sending it to Linus.

In view of the prospective improvements around shutdown mode please
leave the related code as is.

>>>>> - I thought the code read better with a bool parameter, vs a longer
>>>>>      function name.
>>>>>
>>>>> So nothing really critical, but mostly just my aesthetic and preference.
>>>>>
>>>>> Also, I expect that is31fl32xx_software_shutdown() would be inlined, so
>>>>> the conditional check in there is optimized out anyways, and there is no
>>>>> performance penalty. Looking at the disassembly in my ARMv7a build, both
>>>>> of those have indeed happened there.
>>>>
>>>> Not only the size of a binary and the performance, but also code
>>>> readability matters. The function is called only with false argument,
>>>> so I expect that some people may submit patches optimizing this.
>>>>
>>>> Let's avoid the confusion.
>>>
>>> I guess we just have a difference of opinion on which way is more
>>> readable, which is OK. Unless the above explanation causes you to
>>> change your mind, I will remove the 'enable' parameter and add a
>>> "_disable" suffix to both functions. That will also leave the
>>> 	#define IS31FL32XX_SHUTDOWN_SSD_ENABLE  0
>>> constant unused in the code, should that also be removed? Note that
>>> the 3216 specific constant
>>> 	#define IS31FL3216_CONFIG_SSD_ENABLE  BIT(7)
>>> would still be used in is31fl3216_reset().
>>
>> Current version of the function would be required for enabling
>> shutting down the controller from brightness_set op when the
>> count of active sub-LEDs drops to 0.
>
> Agreed. That's a reason I would have for leaving it so even if the
> auto-software-shutdown logic is not implemented at this time. But I
> would understand if you still prefer to only have the current form
> if there is currently code which uses both possible values of the
> 'enable' parameter.

We can stay by the current implementation.

>>>>>>>>> +	return 0;
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
>>>>>>>>> +{
>>>>>>>>> +	const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>>>>>> +	int ret;
>>>>>>>>> +
>>>>>>>>> +	ret = is31fl32xx_reset_regs(priv);
>>>>>>>>> +	if (ret)
>>>>>>>>> +		return ret;
>>>>>>>>> +
>>>>>>>>> +	/*
>>>>>>>>> +	 * Set enable bit for all channels.
>>>>>>>>> +	 * We will control state with PWM registers alone.
>>>>>>>>> +	 */
>>>>>>>>> +	if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
>>>>>>>>> +		u8 value =
>>>>>>>>> +		    GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
>>>>>>>>> +		u8 num_regs = cdef->channels /
>>>>>>>>> +				cdef->enable_bits_per_led_control_register;
>>>>>>>>> +		int i;
>>>>>>>>> +
>>>>>>>>> +		for (i = 0; i < num_regs; i++) {
>>>>>>>>> +			ret = is31fl32xx_write(priv,
>>>>>>>>> +					       cdef->led_control_register_base+i,
>>>>>>>>> +					       value);
>>>>>>>>> +			if (ret)
>>>>>>>>> +				return ret;
>>>>>>>>> +		}
>>>>>>>>> +	}
>>>>>>>>> +
>>>>>>>>> +	ret = is31fl32xx_software_shutdown(priv, false);
>>>>>>>>> +	if (ret)
>>>>>>>>> +		return ret;
>>>>>>>>> +
>>>>>>>>> +	if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
>>>>>>>>> +		ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
>>>>>>>>> +		if (ret)
>>>>>>>>> +			return ret;
>>>>>>>>> +	}
>>>>>>>>> +
>>>>>>>>> +	return 0;
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
>>>>>>>>> +{
>>>>>>>>> +	return sizeof(struct is31fl32xx_priv) +
>>>>>>>>> +		      (sizeof(struct is31fl32xx_led_data) * num_leds);
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
>>>>>>>>> +				     const struct device_node *child,
>>>>>>>>> +				     struct is31fl32xx_led_data *led_data)
>>>>>>>>> +{
>>>>>>>>> +	struct led_classdev *cdev = &led_data->cdev;
>>>>>>>>> +	int ret = 0;
>>>>>>>>> +	u32 reg;
>>>>>>>>> +
>>>>>>>>> +	if (of_property_read_string(child, "label", &cdev->name))
>>>>>>>>> +		cdev->name = child->name;
>>>>>>>>> +
>>>>>>>>> +	ret = of_property_read_u32(child, "reg", &reg);
>>>>>>>>> +	if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
>>>>>>>>> +		dev_err(dev,
>>>>>>>>> +			"Child node %s does not have a valid reg property\n",
>>>>>>>>> +			child->full_name);
>>>>>>>>> +		return -EINVAL;
>>>>>>>>> +	}
>>>>>>>>> +	led_data->channel = reg;
>>>>>>>>> +
>>>>>>>>> +	of_property_read_string(child, "linux,default-trigger",
>>>>>>>>> +				&cdev->default_trigger);
>>>>>>>>> +
>>>>>>>>> +	cdev->brightness_set_blocking = is31fl32xx_brightness_set;
>>>>>>>>> +
>>>>>>>>> +	return 0;
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
>>>>>>>>> +					struct is31fl32xx_priv *priv,
>>>>>>>>> +					u8 channel)
>>>>>>>>> +{
>>>>>>>>> +	size_t i;
>>>>>>>>> +
>>>>>>>>> +	for (i = 0; i < priv->num_leds; i++) {
>>>>>>>>> +		if (priv->leds[i].channel == channel)
>>>>>>>>> +			return &priv->leds[i];
>>>>>>>>> +	}
>>>>>>>>> +
>>>>>>>>> +	return NULL;
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +static int is31fl32xx_parse_dt(struct device *dev,
>>>>>>>>> +			       struct is31fl32xx_priv *priv)
>>>>>>>>> +{
>>>>>>>>> +	struct device_node *child;
>>>>>>>>> +	int ret = 0;
>>>>>>>>> +
>>>>>>>>> +	for_each_child_of_node(dev->of_node, child) {
>>>>>>>>> +		struct is31fl32xx_led_data *led_data =
>>>>>>>>> +			&priv->leds[priv->num_leds];
>>>>>>>>> +		const struct is31fl32xx_led_data *other_led_data;
>>>>>>>>> +
>>>>>>>>> +		led_data->priv = priv;
>>>>>>>>> +
>>>>>>>>> +		ret = is31fl32xx_parse_child_dt(dev, child, led_data);
>>>>>>>>> +		if (ret)
>>>>>>>>> +			goto err;
>>>>>>>>> +
>>>>>>>>> +		/* Detect if channel is already in use by another child */
>>>>>>>>> +		other_led_data = is31fl32xx_find_led_data(priv,
>>>>>>>>> +							  led_data->channel);
>>>>>>>>> +		if (other_led_data) {
>>>>>>>>> +			dev_err(dev,
>>>>>>>>> +				"%s and %s both attempting to use channel %d\n",
>>>>>>>>> +				led_data->cdev.name,
>>>>>>>>> +				other_led_data->cdev.name,
>>>>>>>>> +				led_data->channel);
>>>>>>>>> +			goto err;
>>>>>>>>> +		}
>>>>>>>>> +
>>>>>>>>> +		ret = devm_led_classdev_register(dev, &led_data->cdev);
>>>>>>>>> +		if (ret) {
>>>>>>>>> +			dev_err(dev, "failed to register PWM led for %s: %d\n",
>>>>>>>>> +				led_data->cdev.name, ret);
>>>>>>>>> +			goto err;
>>>>>>>>> +		}
>>>>>>>>> +
>>>>>>>>> +		priv->num_leds++;
>>>>>>>>> +	}
>>>>>>>>> +
>>>>>>>>> +	return 0;
>>>>>>>>> +
>>>>>>>>> +err:
>>>>>>>>> +	of_node_put(child);
>>>>>>>>> +	return ret;
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +static const struct of_device_id of_is31fl31xx_match[] = {
>>>>>>>>> +	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
>>>>>>>>> +	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
>>>>>>>>> +	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
>>>>>>>>> +	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
>>>>>>>>> +	{},
>>>>>>>>> +};
>>>>>>>>> +
>>>>>>>>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
>>>>>>>>> +
>>>>>>>>> +static int is31fl32xx_probe(struct i2c_client *client,
>>>>>>>>> +			    const struct i2c_device_id *id)
>>>>>>>>> +{
>>>>>>>>> +	const struct is31fl32xx_chipdef *cdef;
>>>>>>>>> +	const struct of_device_id *of_dev_id;
>>>>>>>>> +	struct device *dev = &client->dev;
>>>>>>>>> +	struct is31fl32xx_priv *priv;
>>>>>>>>> +	int count;
>>>>>>>>> +	int ret = 0;
>>>>>>>>> +
>>>>>>>>> +	of_dev_id = of_match_device(of_is31fl31xx_match, dev);
>>>>>>>>> +	if (!of_dev_id)
>>>>>>>>> +		return -EINVAL;
>>>>>>>>> +
>>>>>>>>> +	cdef = of_dev_id->data;
>>>>>>>>> +
>>>>>>>>> +	count = of_get_child_count(dev->of_node);
>>>>>>>>> +	if (!count)
>>>>>>>>> +		return -EINVAL;
>>>>>>>>> +
>>>>>>>>> +	priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
>>>>>>>>> +			    GFP_KERNEL);
>>>>>>>>> +	if (!priv)
>>>>>>>>> +		return -ENOMEM;
>>>>>>>>> +
>>>>>>>>> +	priv->client = client;
>>>>>>>>> +	priv->cdef = cdef;
>>>>>>>>> +	i2c_set_clientdata(client, priv);
>>>>>>>>> +
>>>>>>>>> +	ret = is31fl32xx_init_regs(priv);
>>>>>>>>> +	if (ret)
>>>>>>>>> +		return ret;
>>>>>>>>> +
>>>>>>>>> +	ret = is31fl32xx_parse_dt(dev, priv);
>>>>>>>>> +	if (ret)
>>>>>>>>> +		return ret;
>>>>>>>>> +
>>>>>>>>> +	return 0;
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +static int is31fl32xx_remove(struct i2c_client *client)
>>>>>>>>> +{
>>>>>>>>> +	struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
>>>>>>>>> +
>>>>>>>>> +	return is31fl32xx_reset_regs(priv);
>>>>>>>>> +}
>>>>>>>>> +
>>>>>>>>> +/*
>>>>>>>>> + * i2c-core requires that id_table be non-NULL, even though
>>>>>>>>> + * it is not used for DeviceTree based instantiation.
>>>>>>>>> + */
>>>>>>>>> +static const struct i2c_device_id is31fl31xx_id[] = {
>>>>>>>>> +	{},
>>>>>>>>> +};
>>>>>>>>> +
>>>>>>>>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
>>>>>>>>> +
>>>>>>>>> +static struct i2c_driver is31fl32xx_driver = {
>>>>>>>>> +	.driver = {
>>>>>>>>> +		.name	= "is31fl32xx",
>>>>>>>>> +		.of_match_table = of_is31fl31xx_match,
>>>>>>>>> +	},
>>>>>>>>> +	.probe		= is31fl32xx_probe,
>>>>>>>>> +	.remove		= is31fl32xx_remove,
>>>>>>>>> +	.id_table	= is31fl31xx_id,
>>>>>>>>> +};
>>>>>>>>> +
>>>>>>>>> +module_i2c_driver(is31fl32xx_driver);
>>>>>>>>> +
>>>>>>>>> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
>>>>>>>>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
>>>>>>>>> +MODULE_LICENSE("GPL v2");
>>>>>>>>>
>

-- 
Best Regards,
Jacek Anaszewski

Re: [PATCH 4/4] leds: Replace dedicated SN3218 driver with IS31FL32XX driver

[Jacek Anaszewski]

On 03/05/2016 06:34 AM, David Rivshin (Allworx) wrote:
> On Thu, 03 Mar 2016 15:51:36 +0100
> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>
>> Hi David,
>>
>> I'll wait for Tested-by from Stefan before applying this patch.
>> If Stefan will have managed to test your driver with his hardware
>> by the end of this cycle, it will suffice for this patch to contain
>> only leds-is31fl32xx extension part.
>>
>> leds-sn3218 hasn't been merged to mainline yet, so I'd rather
>> remove it when I get Tested-by from Stefan.
>>
>> Otherwise, I will send leds-sn3218 to Linus, and this patch
>> will be applied after being tested, during 4.7 cycle.
>
> Since Stefan has given his Tested-by, do I understand correctly
> that you would prefer to:
>   - revert sn3218 patches 2 and 3 yourself
>   - have me provide v2 patches that would apply ontop of that

Correct.

> If so, should I do as Rob suggests and fold the sn3216/3218 bits into
> the earlier patches, or leave it as a 4th patch? I'd probably prefer
> the former as being an easier workflow (less "rebase -i"ing).

I think that it would be more striking if leds-sn3218 specific
compatible would be added both to the driver and the documentation
as separate patches. Separate commit description would highlight
the reasons for which the same driver supports compatibles with
different vendor prefixes.

> And should I base off v4.5-rc6 at that point, or your for-next minus
> those patches?

I like the latter option. I'll do the same - will drop leds-sn3218
patches and apply yours.

> The only difference should be a 1-line offset in the
> vendor-prefixes.txt hunk. Either way is equally easy for me.
>
>>
>> Thanks,
>> Jacek Anaszewski
>>
>> On 03/03/2016 04:01 AM, David Rivshin (Allworx) wrote:
>>> From: David Rivshin <drivshin@allworx.com>
>>>
>>> Si-En Technology was acquired by ISSI in 2011, and it appears that
>>> the IS31FL3218/IS31FL3216 are just rebranded SN3218/SN3216 devices.
>>> As the IS31FL32XX driver already handles the *3218 devices, there
>>> is no longer a need for the dedicated SN3218 driver.
>>>
>>> Add the "sn,sn3218" and "sn,sn3216" compatible strings into the
>>> IS31FL32XX driver and binding documentation, and remove the
>>> leds-sn3218 driver.
>>>
>>> Datasheets:
>>>       IS31FL3218: http://www.issi.com/WW/pdf/31FL3218.pdf
>>>       SN3218:     http://www.si-en.com/uploadpdf/s2011517171720.pdf
>>>
>>>       IS31FL3216: http://www.issi.com/WW/pdf/31FL3216.pdf
>>>       SN3216;     http://www.si-en.com/uploadpdf/SN3216201152410148.pdf
>>>
>>> Signed-off-by: David Rivshin <drivshin@allworx.com>
>>> ---
>>>
>>> Note that the leds-sn3218 binding use a 0-based 'reg' property, while
>>> the leds-is31fl32xx binding uses a 1-based 'reg' property. This seemed
>>> to be the preferred binding based on [1]. Since leds-sn3216 has not been
>>> in a released kernel, there is are no backwards-compatibility concerns.
>>>
>>> Changes from RFC:
>>>    new
>>>
>>> [1] http://www.spinics.net/lists/linux-leds/msg05589.html
>>>
>>>    .../devicetree/bindings/leds/leds-is31fl32xx.txt   |   9 +-
>>>    .../devicetree/bindings/leds/leds-sn3218.txt       |  41 ---
>>>    drivers/leds/Kconfig                               |  18 +-
>>>    drivers/leds/Makefile                              |   1 -
>>>    drivers/leds/leds-is31fl32xx.c                     |   6 +-
>>>    drivers/leds/leds-sn3218.c                         | 306 ---------------------
>>>    6 files changed, 14 insertions(+), 367 deletions(-)
>>>    delete mode 100644 Documentation/devicetree/bindings/leds/leds-sn3218.txt
>>>    delete mode 100644 drivers/leds/leds-sn3218.c
>>>
>>> diff --git a/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
>>> index 539df2e..c59eb1a 100644
>>> --- a/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
>>> +++ b/Documentation/devicetree/bindings/leds/leds-is31fl32xx.txt
>>> @@ -1,6 +1,6 @@
>>> -Binding for ISSI IS31FL32xx LED Drivers
>>> +Binding for ISSI IS31FL32xx and Si-En SN32xx LED Drivers
>>>
>>> -The IS31FL32xx family of LED drivers are I2C devices with multiple
>>> +The IS31FL32xx/SN32xx family of LED drivers are I2C devices with multiple
>>>    constant-current channels, each with independent 256-level PWM control.
>>>    Each LED is represented as a sub-node of the device.
>>>
>>> @@ -10,6 +10,8 @@ Required properties:
>>>    	issi,is31fl3235
>>>    	issi,is31fl3218
>>>    	issi,is31fl3216
>>> +	si-en,sn3218
>>> +	si-en,sn3216
>>>    - reg: I2C slave address
>>>    - address-cells : must be 1
>>>    - size-cells : must be 0
>>> @@ -45,5 +47,6 @@ leds: is31fl3236@3c {
>>>    	};
>>>    };
>>>
>>> -For more product information please see the link below:
>>> +For more product information please see the links below:
>>>    http://www.issi.com/US/product-analog-fxled-driver.shtml
>>> +http://www.si-en.com/product.asp?parentid=890
>>> diff --git a/Documentation/devicetree/bindings/leds/leds-sn3218.txt b/Documentation/devicetree/bindings/leds/leds-sn3218.txt
>>> deleted file mode 100644
>>> index 19cbf57..0000000
>>> --- a/Documentation/devicetree/bindings/leds/leds-sn3218.txt
>>> +++ /dev/null
>>> @@ -1,41 +0,0 @@
>>> -* Si-En Technology - SN3218 18-Channel LED Driver
>>> -
>>> -Required properties:
>>> -- compatible :
>>> -	"si-en,sn3218"
>>> -- address-cells : must be 1
>>> -- size-cells : must be 0
>>> -- reg : I2C slave address, typically 0x54
>>> -
>>> -There must be at least 1 LED which is represented as a sub-node
>>> -of the sn3218 device.
>>> -
>>> -LED sub-node properties:
>>> -- label : (optional) see Documentation/devicetree/bindings/leds/common.txt
>>> -- reg : number of LED line (could be from 0 to 17)
>>> -- linux,default-trigger : (optional)
>>> -   see Documentation/devicetree/bindings/leds/common.txt
>>> -
>>> -Example:
>>> -
>>> -sn3218: led-controller@54 {
>>> -	compatible = "si-en,sn3218";
>>> -	#address-cells = <1>;
>>> -	#size-cells = <0>;
>>> -	reg = <0x54>;
>>> -
>>> -	led@0 {
>>> -		label = "led1";
>>> -		reg = <0x0>;
>>> -	};
>>> -
>>> -	led@1 {
>>> -		label = "led2";
>>> -		reg = <0x1>;
>>> -	};
>>> -
>>> -	led@2 {
>>> -		label = "led3";
>>> -		reg = <0x2>;
>>> -	};
>>> -};
>>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
>>> index 9c63ba4..1f64151 100644
>>> --- a/drivers/leds/Kconfig
>>> +++ b/drivers/leds/Kconfig
>>> @@ -568,25 +568,13 @@ config LEDS_SEAD3
>>>    	  This driver can also be built as a module. If so the module
>>>    	  will be called leds-sead3.
>>>
>>> -config LEDS_SN3218
>>> -	tristate "LED support for Si-En SN3218 I2C chip"
>>> -	depends on LEDS_CLASS && I2C
>>> -	depends on OF
>>> -	select REGMAP_I2C
>>> -	help
>>> -	  This option enables support for the Si-EN SN3218 LED driver
>>> -	  connected through I2C. Say Y to enable support for the SN3218 LED.
>>> -
>>> -	  This driver can also be built as a module. If so the module
>>> -	  will be called leds-sn3218.
>>> -
>>>    config LEDS_IS31FL32XX
>>>    	tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
>>>    	depends on LEDS_CLASS && I2C && OF
>>>    	help
>>> -	  Say Y here to include support for ISSI IS31FL32XX LED controllers.
>>> -	  They are I2C devices with multiple constant-current channels, each
>>> -	  with independent 256-level PWM control.
>>> +	  Say Y here to include support for ISSI IS31FL32XX and Si-En SN32xx
>>> +	  LED controllers. They are I2C devices with multiple constant-current
>>> +	  channels, each with independent 256-level PWM control.
>>>
>>>    comment "LED driver for blink(1) USB RGB LED is under Special HID drivers (HID_THINGM)"
>>>
>>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
>>> index 3fdf313..cb2013d 100644
>>> --- a/drivers/leds/Makefile
>>> +++ b/drivers/leds/Makefile
>>> @@ -66,7 +66,6 @@ obj-$(CONFIG_LEDS_MENF21BMC)		+= leds-menf21bmc.o
>>>    obj-$(CONFIG_LEDS_KTD2692)		+= leds-ktd2692.o
>>>    obj-$(CONFIG_LEDS_POWERNV)		+= leds-powernv.o
>>>    obj-$(CONFIG_LEDS_SEAD3)		+= leds-sead3.o
>>> -obj-$(CONFIG_LEDS_SN3218)		+= leds-sn3218.o
>>>    obj-$(CONFIG_LEDS_IS31FL32XX)		+= leds-is31fl32xx.o
>>>
>>>    # LED SPI Drivers
>>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
>>> index 49818f0..ec3f541 100644
>>> --- a/drivers/leds/leds-is31fl32xx.c
>>> +++ b/drivers/leds/leds-is31fl32xx.c
>>> @@ -10,7 +10,9 @@
>>>     * it under the terms of the GNU General Public License version 2 as
>>>     * published by the Free Software Foundation.
>>>     *
>>> - * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>> + * Datasheets:
>>> + *   http://www.issi.com/US/product-analog-fxled-driver.shtml
>>> + *   http://www.si-en.com/product.asp?parentid=890
>>>     */
>>>
>>>    #include <linux/err.h>
>>> @@ -425,7 +427,9 @@ static const struct of_device_id of_is31fl31xx_match[] = {
>>>    	{ .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
>>>    	{ .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
>>>    	{ .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
>>> +	{ .compatible = "si-en,sn3218",    .data = &is31fl3218_cdef, },
>>>    	{ .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
>>> +	{ .compatible = "si-en,sn3216",    .data = &is31fl3216_cdef, },
>>>    	{},
>>>    };
>>>
>>> diff --git a/drivers/leds/leds-sn3218.c b/drivers/leds/leds-sn3218.c
>>> deleted file mode 100644
>>> index dcc2581..0000000
>>> --- a/drivers/leds/leds-sn3218.c
>>> +++ /dev/null
>>> @@ -1,306 +0,0 @@
>>> -/*
>>> - * Si-En SN3218 18 Channel LED Driver
>>> - *
>>> - * Copyright (C) 2016 Stefan Wahren <stefan.wahren@i2se.com>
>>> - *
>>> - * Based on leds-pca963x.c
>>> - *
>>> - * 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.
>>> - *
>>> - * Datasheet: http://www.si-en.com/uploadpdf/s2011517171720.pdf
>>> - *
>>> - */
>>> -
>>> -#include <linux/err.h>
>>> -#include <linux/i2c.h>
>>> -#include <linux/leds.h>
>>> -#include <linux/module.h>
>>> -#include <linux/of.h>
>>> -#include <linux/regmap.h>
>>> -#include <linux/slab.h>
>>> -
>>> -#define SN3218_MODE		0x00
>>> -#define SN3218_PWM_1		0x01
>>> -#define SN3218_PWM_2		0x02
>>> -#define SN3218_PWM_3		0x03
>>> -#define SN3218_PWM_4		0x04
>>> -#define SN3218_PWM_5		0x05
>>> -#define SN3218_PWM_6		0x06
>>> -#define SN3218_PWM_7		0x07
>>> -#define SN3218_PWM_8		0x08
>>> -#define SN3218_PWM_9		0x09
>>> -#define SN3218_PWM_10		0x0a
>>> -#define SN3218_PWM_11		0x0b
>>> -#define SN3218_PWM_12		0x0c
>>> -#define SN3218_PWM_13		0x0d
>>> -#define SN3218_PWM_14		0x0e
>>> -#define SN3218_PWM_15		0x0f
>>> -#define SN3218_PWM_16		0x10
>>> -#define SN3218_PWM_17		0x11
>>> -#define SN3218_PWM_18		0x12
>>> -#define SN3218_LED_1_6		0x13
>>> -#define SN3218_LED_7_12		0x14
>>> -#define SN3218_LED_13_18	0x15
>>> -#define SN3218_UPDATE		0x16	/* applies to reg 0x01 .. 0x15 */
>>> -#define SN3218_RESET		0x17
>>> -
>>> -#define SN3218_LED_MASK		0x3f
>>> -#define SN3218_LED_ON		0x01
>>> -#define SN3218_LED_OFF		0x00
>>> -
>>> -#define SN3218_MODE_SHUTDOWN	0x00
>>> -#define SN3218_MODE_NORMAL	0x01
>>> -
>>> -#define NUM_LEDS		18
>>> -
>>> -struct sn3218_led;
>>> -
>>> -/**
>>> - * struct sn3218 -
>>> - * @client - Pointer to the I2C client
>>> - * @leds - Pointer to the individual LEDs
>>> - * @num_leds - Actual number of LEDs
>>> -**/
>>> -struct sn3218 {
>>> -	struct i2c_client *client;
>>> -	struct regmap *regmap;
>>> -	struct sn3218_led *leds;
>>> -	int num_leds;
>>> -};
>>> -
>>> -/**
>>> - * struct sn3218_led -
>>> - * @chip - Pointer to the container
>>> - * @led_cdev - led class device pointer
>>> - * @led_num - LED index ( 0 .. 17 )
>>> -**/
>>> -struct sn3218_led {
>>> -	struct sn3218 *chip;
>>> -	struct led_classdev led_cdev;
>>> -	int led_num;
>>> -};
>>> -
>>> -static int sn3218_led_set(struct led_classdev *led_cdev,
>>> -			  enum led_brightness brightness)
>>> -{
>>> -	struct sn3218_led *led =
>>> -			container_of(led_cdev, struct sn3218_led, led_cdev);
>>> -	struct regmap *regmap = led->chip->regmap;
>>> -	u8 bank = led->led_num / 6;
>>> -	u8 mask = 0x1 << (led->led_num % 6);
>>> -	u8 val;
>>> -	int ret;
>>> -
>>> -	if (brightness == LED_OFF)
>>> -		val = 0;
>>> -	else
>>> -		val = mask;
>>> -
>>> -	ret = regmap_update_bits(regmap, SN3218_LED_1_6 + bank, mask, val);
>>> -	if (ret < 0)
>>> -		return ret;
>>> -
>>> -	if (brightness > LED_OFF) {
>>> -		ret = regmap_write(regmap, SN3218_PWM_1 + led->led_num,
>>> -				   brightness);
>>> -		if (ret < 0)
>>> -			return ret;
>>> -	}
>>> -
>>> -	ret = regmap_write(regmap, SN3218_UPDATE, 0xff);
>>> -
>>> -	return ret;
>>> -}
>>> -
>>> -static void sn3218_led_init(struct sn3218 *sn3218, struct device_node *node,
>>> -			    u32 reg)
>>> -{
>>> -	struct sn3218_led *leds = sn3218->leds;
>>> -	struct led_classdev *cdev = &leds[reg].led_cdev;
>>> -
>>> -	leds[reg].led_num = reg;
>>> -	leds[reg].chip = sn3218;
>>> -
>>> -	if (of_property_read_string(node, "label", &cdev->name))
>>> -		cdev->name = node->name;
>>> -
>>> -	of_property_read_string(node, "linux,default-trigger",
>>> -				&cdev->default_trigger);
>>> -
>>> -	cdev->brightness_set_blocking = sn3218_led_set;
>>> -}
>>> -
>>> -static const struct reg_default sn3218_reg_defs[] = {
>>> -	{ SN3218_MODE, 0x00},
>>> -	{ SN3218_PWM_1, 0x00},
>>> -	{ SN3218_PWM_2, 0x00},
>>> -	{ SN3218_PWM_3, 0x00},
>>> -	{ SN3218_PWM_4, 0x00},
>>> -	{ SN3218_PWM_5, 0x00},
>>> -	{ SN3218_PWM_6, 0x00},
>>> -	{ SN3218_PWM_7, 0x00},
>>> -	{ SN3218_PWM_8, 0x00},
>>> -	{ SN3218_PWM_9, 0x00},
>>> -	{ SN3218_PWM_10, 0x00},
>>> -	{ SN3218_PWM_11, 0x00},
>>> -	{ SN3218_PWM_12, 0x00},
>>> -	{ SN3218_PWM_13, 0x00},
>>> -	{ SN3218_PWM_14, 0x00},
>>> -	{ SN3218_PWM_15, 0x00},
>>> -	{ SN3218_PWM_16, 0x00},
>>> -	{ SN3218_PWM_17, 0x00},
>>> -	{ SN3218_PWM_18, 0x00},
>>> -	{ SN3218_LED_1_6, 0x00},
>>> -	{ SN3218_LED_7_12, 0x00},
>>> -	{ SN3218_LED_13_18, 0x00},
>>> -	{ SN3218_UPDATE, 0x00},
>>> -	{ SN3218_RESET, 0x00},
>>> -};
>>> -
>>> -static const struct regmap_config sn3218_regmap_config = {
>>> -	.reg_bits = 8,
>>> -	.val_bits = 8,
>>> -
>>> -	.max_register = SN3218_RESET,
>>> -	.reg_defaults = sn3218_reg_defs,
>>> -	.num_reg_defaults = ARRAY_SIZE(sn3218_reg_defs),
>>> -	.cache_type = REGCACHE_RBTREE,
>>> -};
>>> -
>>> -static int sn3218_init(struct i2c_client *client, struct sn3218 *sn3218)
>>> -{
>>> -	struct device_node *np = client->dev.of_node, *child;
>>> -	struct sn3218_led *leds;
>>> -	int ret, count;
>>> -
>>> -	count = of_get_child_count(np);
>>> -	if (!count)
>>> -		return -ENODEV;
>>> -
>>> -	if (count > NUM_LEDS) {
>>> -		dev_err(&client->dev, "Invalid LED count %d\n", count);
>>> -		return -EINVAL;
>>> -	}
>>> -
>>> -	leds = devm_kzalloc(&client->dev, count * sizeof(*leds), GFP_KERNEL);
>>> -	if (!leds)
>>> -		return -ENOMEM;
>>> -
>>> -	sn3218->leds = leds;
>>> -	sn3218->num_leds = count;
>>> -	sn3218->client = client;
>>> -
>>> -	sn3218->regmap = devm_regmap_init_i2c(client, &sn3218_regmap_config);
>>> -	if (IS_ERR(sn3218->regmap)) {
>>> -		ret = PTR_ERR(sn3218->regmap);
>>> -		dev_err(&client->dev, "Failed to allocate register map: %d\n",
>>> -			ret);
>>> -		return ret;
>>> -	}
>>> -
>>> -	for_each_child_of_node(np, child) {
>>> -		u32 reg;
>>> -
>>> -		ret = of_property_read_u32(child, "reg", &reg);
>>> -		if (ret)
>>> -			goto fail;
>>> -
>>> -		if (reg >= count) {
>>> -			dev_err(&client->dev, "Invalid LED (%u >= %d)\n", reg,
>>> -				count);
>>> -			ret = -EINVAL;
>>> -			goto fail;
>>> -		}
>>> -
>>> -		sn3218_led_init(sn3218, child, reg);
>>> -	}
>>> -
>>> -	return 0;
>>> -
>>> -fail:
>>> -	of_node_put(child);
>>> -	return ret;
>>> -}
>>> -
>>> -static int sn3218_probe(struct i2c_client *client,
>>> -			const struct i2c_device_id *id)
>>> -{
>>> -	struct sn3218 *sn3218;
>>> -	struct sn3218_led *leds;
>>> -	struct device *dev = &client->dev;
>>> -	int i, ret;
>>> -
>>> -	sn3218 = devm_kzalloc(dev, sizeof(*sn3218), GFP_KERNEL);
>>> -	if (!sn3218)
>>> -		return -ENOMEM;
>>> -
>>> -	ret = sn3218_init(client, sn3218);
>>> -	if (ret)
>>> -		return ret;
>>> -
>>> -	i2c_set_clientdata(client, sn3218);
>>> -	leds = sn3218->leds;
>>> -
>>> -	/*
>>> -	 * Since the chip is write-only we need to reset him into
>>> -	 * a defined state (all LEDs off).
>>> -	 */
>>> -	ret = regmap_write(sn3218->regmap, SN3218_RESET, 0xff);
>>> -	if (ret)
>>> -		return ret;
>>> -
>>> -	for (i = 0; i < sn3218->num_leds; i++) {
>>> -		ret = devm_led_classdev_register(dev, &leds[i].led_cdev);
>>> -		if (ret < 0)
>>> -			return ret;
>>> -	}
>>> -
>>> -	return regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_NORMAL);
>>> -}
>>> -
>>> -static int sn3218_remove(struct i2c_client *client)
>>> -{
>>> -	struct sn3218 *sn3218 = i2c_get_clientdata(client);
>>> -
>>> -	regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_SHUTDOWN);
>>> -
>>> -	return 0;
>>> -}
>>> -
>>> -static void sn3218_shutdown(struct i2c_client *client)
>>> -{
>>> -	struct sn3218 *sn3218 = i2c_get_clientdata(client);
>>> -
>>> -	regmap_write(sn3218->regmap, SN3218_MODE, SN3218_MODE_SHUTDOWN);
>>> -}
>>> -
>>> -static const struct i2c_device_id sn3218_id[] = {
>>> -	{ "sn3218", 0 },
>>> -	{ }
>>> -};
>>> -MODULE_DEVICE_TABLE(i2c, sn3218_id);
>>> -
>>> -static const struct of_device_id of_sn3218_match[] = {
>>> -	{ .compatible = "si-en,sn3218", },
>>> -	{},
>>> -};
>>> -MODULE_DEVICE_TABLE(of, of_sn3218_match);
>>> -
>>> -static struct i2c_driver sn3218_driver = {
>>> -	.driver = {
>>> -		.name	= "leds-sn3218",
>>> -		.of_match_table = of_match_ptr(of_sn3218_match),
>>> -	},
>>> -	.probe	= sn3218_probe,
>>> -	.remove	= sn3218_remove,
>>> -	.shutdown = sn3218_shutdown,
>>> -	.id_table = sn3218_id,
>>> -};
>>> -
>>> -module_i2c_driver(sn3218_driver);
>>> -
>>> -MODULE_DESCRIPTION("Si-En SN3218 LED Driver");
>>> -MODULE_AUTHOR("Stefan Wahren <stefan.wahren@i2se.com>");
>>> -MODULE_LICENSE("GPL v2");
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-leds" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

-- 
Best Regards,
Jacek Anaszewski

Re: [PATCH 3/4] leds: Add driver for the ISSI IS31FL32xx family of LED controllers

[Jacek Anaszewski]

On 03/05/2016 07:12 AM, David Rivshin (Allworx) wrote:
> On Fri, 04 Mar 2016 16:38:01 +0100
> Jacek Anaszewski <j.anaszewski@samsung.com> wrote:
>
>> On 03/04/2016 03:27 PM, David Rivshin (Allworx) wrote:
>>> (Stefan, sorry for the duplicate, I just realized that I originally
>>> replied only to you by accident).
>>>
>>> On Thu, 3 Mar 2016 19:13:03 +0100 (CET)
>>> Stefan Wahren <stefan.wahren@i2se.com> wrote:
>>>
>>>> Hi David,
>>>>
>>>> i will test the driver on weekend. Some comments below
>>>
>>> Great, thanks very much.
>>>
>>>>> "David Rivshin (Allworx)" <drivshin.allworx@gmail.com> hat am 3. März 2016 um
>>>>> 04:01 geschrieben:
>>>>>
>>>>>
>>>>> From: David Rivshin <drivshin@allworx.com>
>>>>>
>>>>> The IS31FL32xx family of LED controllers are I2C devices with multiple
>>>>> constant-current channels, each with independent 256-level PWM control.
>>>>>
>>>>> Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>>>>
>>>>> This has been tested on the IS31FL3236 and IS31FL3216, on an ARM
>>>>> (TI am335x) platform.
>>>>>
>>>>> The programming paradigm of these devices is similar in the following
>>>>> ways:
>>>>> - All registers are 8 bit
>>>>> - All LED control registers are write-only
>>>>> - Each LED channel has a PWM register (0-255)
>>>>> - PWM register writes are shadowed until an Update register is poked
>>>>> - All have a concept of Software Shutdown, which disables output
>>>>>
>>>>> However, there are some differences in devices:
>>>>> - 3236/3235 have a separate Control register for each LED,
>>>>> (3218/3216 pack the enable bits into fewer registers)
>>>>> - 3236/3235 have a per-channel current divisor setting
>>>>> - 3236/3235 have a Global Control register that can turn off all LEDs
>>>>> - 3216 is unique in a number of ways
>>>>> - OUT9-OUT16 can be configured as GPIOs instead of LED controls
>>>>> - LEDs can be programmed with an 8-frame animation, with
>>>>> programmable delay between frames
>>>>> - LEDs can be modulated by an input audio signal
>>>>> - Max output current can be adjusted from 1/4 to 2x globally
>>>>> - Has a Configuration register instead of a Shutdown register
>>>>>
>>>>> This driver currently only supports the base PWM control function
>>>>> of these devices. The following features of these devices are not
>>>>> implemented, although it should be possible to add them in the future:
>>>>> - All devices are capable of going into a lower-power "software
>>>>> shutdown" mode.
>>>>> - The is31fl3236 and is31fl3235 can reduce the max output current
>>>>> per-channel with a divisor of 1, 2, 3, or 4.
>>>>> - The is31fl3216 can use some LED channels as GPIOs instead.
>>>>> - The is31fl3216 can animate LEDs in hardware.
>>>>> - The is31fl3216 can modulate LEDs according to an audio input.
>>>>> - The is31fl3216 can reduce/increase max output current globally.
>>>>>
>>>>> Signed-off-by: David Rivshin <drivshin@allworx.com>
>>>>> ---
>>>>>
>>>>> You may see two instances of this warning:
>>>>> "passing argument 1 of 'of_property_read_string' discards 'const'
>>>>> qualifier from pointer target type"
>>>>> That is a result of of_property_read_string() taking a non-const
>>>>> struct device_node pointer parameter. I have separately submitted a
>>>>> patch to fix that [1], and a few related functions which had the same
>>>>> issue. I'm hoping that will get into linux-next before this does, so
>>>>> that the warnings never show up there.
>>>>>
>>>>> Changes from RFC:
>>>>> - Removed max-brightness DT property.
>>>>> - Refer to these devices as "LED controllers" in Kconfig.
>>>>> - Removed redundant last sentence from Kconfig entry
>>>>> - Removed unnecessary debug code.
>>>>> - Do not set led_classdev.brightness to 0 explicitly, as it is
>>>>> already initialized to 0 by devm_kzalloc().
>>>>> - Used of_property_read_string() instead of of_get_property().
>>>>> - Fail immediately on DT parsing error in a child node, rather than
>>>>> continuing on with the non-faulty ones.
>>>>> - Added additional comments for some things that might be non-obvious.
>>>>> - Added constants for the location of the SSD bit in the SHUTDOWN
>>>>> register, and the 3216's CONFIG register.
>>>>> - Added special sw_shutdown_func for the 3216 device, as that bit
>>>>> is in a different register, at a different position, and has reverse
>>>>> polarity compared to all the other devices.
>>>>> - Refactored is31fl32xx_init_regs() to separate out some logic into
>>>>> is31fl32xx_reset_regs() and is31fl32xx_software_shutdown().
>>>>>
>>>>> [1] https://lkml.org/lkml/2016/3/2/746
>>>>>
>>>>> drivers/leds/Kconfig | 8 +
>>>>> drivers/leds/Makefile | 1 +
>>>>> drivers/leds/leds-is31fl32xx.c | 505 +++++++++++++++++++++++++++++++++++++++++
>>>>> 3 files changed, 514 insertions(+)
>>>>> create mode 100644 drivers/leds/leds-is31fl32xx.c
>>>>>
>>>>> diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
>>>>> index 1034696..9c63ba4 100644
>>>>> --- a/drivers/leds/Kconfig
>>>>> +++ b/drivers/leds/Kconfig
>>>>> @@ -580,6 +580,14 @@ config LEDS_SN3218
>>>>> This driver can also be built as a module. If so the module
>>>>> will be called leds-sn3218.
>>>>>
>>>>> +config LEDS_IS31FL32XX
>>>>> + tristate "LED support for ISSI IS31FL32XX I2C LED controller family"
>>>>> + depends on LEDS_CLASS && I2C && OF
>>>>> + help
>>>>> + Say Y here to include support for ISSI IS31FL32XX LED controllers.
>>>>> + They are I2C devices with multiple constant-current channels, each
>>>>> + with independent 256-level PWM control.
>>>>
>>>> Is it worth to mention the module name here?
>>>
>>> I noticed that some do and some don't. I don't mind adding it, but it
>>> also seemed like it would be obvious, and therefore unnecessary.
>>>
>>> Jacek, which do you prefer?
>>
>> I agree - it's obvious, we can skip it.
>>
>>>>> +
>>>>> comment "LED driver for blink(1) USB RGB LED is under Special HID drivers
>>>>> (HID_THINGM)"
>>>>>
>>>>> config LEDS_BLINKM
>>>>> diff --git a/drivers/leds/Makefile b/drivers/leds/Makefile
>>>>> index 89c9b6f..3fdf313 100644
>>>>> --- a/drivers/leds/Makefile
>>>>> +++ b/drivers/leds/Makefile
>>>>> @@ -67,6 +67,7 @@ obj-$(CONFIG_LEDS_KTD2692) += leds-ktd2692.o
>>>>> obj-$(CONFIG_LEDS_POWERNV) += leds-powernv.o
>>>>> obj-$(CONFIG_LEDS_SEAD3) += leds-sead3.o
>>>>> obj-$(CONFIG_LEDS_SN3218) += leds-sn3218.o
>>>>> +obj-$(CONFIG_LEDS_IS31FL32XX) += leds-is31fl32xx.o
>>>>>
>>>>> # LED SPI Drivers
>>>>> obj-$(CONFIG_LEDS_DAC124S085) += leds-dac124s085.o
>>>>> diff --git a/drivers/leds/leds-is31fl32xx.c b/drivers/leds/leds-is31fl32xx.c
>>>>> new file mode 100644
>>>>> index 0000000..49818f0
>>>>> --- /dev/null
>>>>> +++ b/drivers/leds/leds-is31fl32xx.c
>>>>> @@ -0,0 +1,505 @@
>>>>> +/*
>>>>> + * linux/drivers/leds-is31fl32xx.c
>>>>
>>>> I think this is unnecessary.
>>>
>>> I tend to agree. I think I used leds-pwm.c as a template, and that had
>>> such a comment. I assumed it was coding-style and kept it, but now I see
>>> that only a minority of led drivers have it. If I do another spin for
>>> any reason I'll remove it.
>>>
>>>>> + *
>>>>> + * Driver for ISSI IS31FL32xx family of I2C LED controllers
>>>>> + *
>>>>> + * Copyright 2015 Allworx Corp.
>>>>> + *
>>>>> + *
>>>>> + * 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.
>>>>> + *
>>>>> + * Datasheets: http://www.issi.com/US/product-analog-fxled-driver.shtml
>>>>> + */
>>>>> +
>>>>
>>>> Shouldn't we include <linux/device.h> here?
>>>
>>> Good catch. I was getting that via i2c.h, but since struct device is
>>> referenced explicitly in a few places, device.h should probably be
>>> included directly.
>>
>> linux/device.h is included from linux/leds.h
>
> This is true, but SubmitChecklist says the following on this topic:
> 1: If you use a facility then #include the file that defines/declares
>     that facility.  Don't depend on other header files pulling in ones
>     that you use.

I'm not sure why declarations are mentioned here. If a type is used
only as an opaque pointer, a declaration suffices. It reduces the number
of build dependencies.

> (Oddly CodingStyle is silent on the topic, which is where I would have
> thought such a thing would be documented.)
>
> I interpreted that to mean that because 'struct device*' appears in
> this file, it should include <linux/device.h> directly.

The header is required if full information about the type implementation
is required, i.e. there are defined variables of the type or the code
refers to the type properties (e.g. dev->of_node).

linux/leds.h could compile without including linux/device.h, if only
it declared struct device, struct attribute_group and
struct device_attribute types.

It might have been that that this include directive has been added
intentionally for the LED class drivers not to be forced to include
it on their own, and indeed most of them don't do it.

Nothing prevents us from sticking to the SubmitChecklist though.

>>>>> +#include <linux/err.h>
>>>>> +#include <linux/i2c.h>
>>>>> +#include <linux/kernel.h>
>>>>> +#include <linux/leds.h>
>>>>> +#include <linux/module.h>
>>>>> +#include <linux/of_platform.h>
>>>>> +
>>>>> +/* Used to indicate a device has no such register */
>>>>> +#define IS31FL32XX_REG_NONE 0xFF
>>>>> +
>>>>> +/* Software Shutdown bit in Shutdown Register */
>>>>> +#define IS31FL32XX_SHUTDOWN_SSD_ENABLE 0
>>>>> +#define IS31FL32XX_SHUTDOWN_SSD_DISABLE BIT(0)
>>>>> +
>>>>> +/* IS31FL3216 has a number of unique registers */
>>>>> +#define IS31FL3216_CONFIG_REG 0x00
>>>>> +#define IS31FL3216_LIGHTING_EFFECT_REG 0x03
>>>>> +#define IS31FL3216_CHANNEL_CONFIG_REG 0x04
>>>>> +
>>>>> +/* Software Shutdown bit in 3216 Config Register */
>>>>> +#define IS31FL3216_CONFIG_SSD_ENABLE BIT(7)
>>>>> +#define IS31FL3216_CONFIG_SSD_DISABLE 0
>>>>> +
>>>>> +struct is31fl32xx_priv;
>>>>> +struct is31fl32xx_led_data {
>>>>> + struct led_classdev cdev;
>>>>> + u8 channel; /* 1-based, max priv->cdef->channels */
>>>>> + struct is31fl32xx_priv *priv;
>>>>> +};
>>>>> +
>>>>> +struct is31fl32xx_priv {
>>>>> + const struct is31fl32xx_chipdef *cdef;
>>>>> + struct i2c_client *client;
>>>>> + unsigned int num_leds;
>>>>> + struct is31fl32xx_led_data leds[0];
>>>>> +};
>>>>> +
>>>>> +/**
>>>>> + * struct is31fl32xx_chipdef - chip-specific attributes
>>>>> + * @channels : Number of LED channels
>>>>> + * @shutdown_reg : address of Shutdown register (optional)
>>>>> + * @pwm_update_reg : address of PWM Update register
>>>>> + * @global_control_reg : address of Global Control register (optional)
>>>>> + * @reset_reg : address of Reset register (optional)
>>>>> + * @pwm_register_base : address of first PWM register
>>>>> + * @pwm_registers_reversed: : true if PWM registers count down instead of up
>>>>> + * @led_control_register_base : address of first LED control register
>>>>> (optional)
>>>>> + * @enable_bits_per_led_control_register: number of LEDs enable bits in each
>>>>> + * @reset_func: : pointer to reset function
>>>>> + *
>>>>> + * For all optional register addresses, the sentinel value
>>>>> %IS31FL32XX_REG_NONE
>>>>> + * indicates that this chip has no such register.
>>>>> + *
>>>>> + * If non-NULL, @reset_func will be called during probing to set all
>>>>> + * necessary registers to a known initialization state. This is needed
>>>>> + * for chips that do not have a @reset_reg.
>>>>> + *
>>>>> + * @enable_bits_per_led_control_register must be >=1 if
>>>>> + * @led_control_register_base != %IS31FL32XX_REG_NONE.
>>>>> + */
>>>>> +struct is31fl32xx_chipdef {
>>>>> + u8 channels;
>>>>> + u8 shutdown_reg;
>>>>> + u8 pwm_update_reg;
>>>>> + u8 global_control_reg;
>>>>> + u8 reset_reg;
>>>>> + u8 pwm_register_base;
>>>>> + bool pwm_registers_reversed;
>>>>> + u8 led_control_register_base;
>>>>> + u8 enable_bits_per_led_control_register;
>>>>> + int (*reset_func)(struct is31fl32xx_priv *priv);
>>>>> + int (*sw_shutdown_func)(struct is31fl32xx_priv *priv, bool enable);
>>>>> +};
>>>>> +
>>>>> +static const struct is31fl32xx_chipdef is31fl3236_cdef = {
>>>>> + .channels = 36,
>>>>> + .shutdown_reg = 0x00,
>>>>> + .pwm_update_reg = 0x25,
>>>>> + .global_control_reg = 0x4a,
>>>>> + .reset_reg = 0x4f,
>>>>> + .pwm_register_base = 0x01,
>>>>> + .led_control_register_base = 0x26,
>>>>> + .enable_bits_per_led_control_register = 1,
>>>>> +};
>>>>> +
>>>>> +static const struct is31fl32xx_chipdef is31fl3235_cdef = {
>>>>> + .channels = 28,
>>>>> + .shutdown_reg = 0x00,
>>>>> + .pwm_update_reg = 0x25,
>>>>> + .global_control_reg = 0x4a,
>>>>> + .reset_reg = 0x4f,
>>>>> + .pwm_register_base = 0x05,
>>>>> + .led_control_register_base = 0x2a,
>>>>> + .enable_bits_per_led_control_register = 1,
>>>>> +};
>>>>> +
>>>>> +static const struct is31fl32xx_chipdef is31fl3218_cdef = {
>>>>> + .channels = 18,
>>>>> + .shutdown_reg = 0x00,
>>>>> + .pwm_update_reg = 0x16,
>>>>> + .global_control_reg = IS31FL32XX_REG_NONE,
>>>>> + .reset_reg = 0x17,
>>>>> + .pwm_register_base = 0x01,
>>>>> + .led_control_register_base = 0x13,
>>>>> + .enable_bits_per_led_control_register = 6,
>>>>> +};
>>>>> +
>>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv);
>>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>>>> + bool enable);
>>>>> +static const struct is31fl32xx_chipdef is31fl3216_cdef = {
>>>>> + .channels = 16,
>>>>> + .shutdown_reg = IS31FL32XX_REG_NONE,
>>>>> + .pwm_update_reg = 0xB0,
>>>>> + .global_control_reg = IS31FL32XX_REG_NONE,
>>>>> + .reset_reg = IS31FL32XX_REG_NONE,
>>>>> + .pwm_register_base = 0x10,
>>>>> + .pwm_registers_reversed = true,
>>>>> + .led_control_register_base = 0x01,
>>>>> + .enable_bits_per_led_control_register = 8,
>>>>> + .reset_func = is31fl3216_reset,
>>>>> + .sw_shutdown_func = is31fl3216_software_shutdown,
>>>>> +};
>>>>> +
>>>>> +static int is31fl32xx_write(struct is31fl32xx_priv *priv, u8 reg, u8 val)
>>>>> +{
>>>>> + int ret;
>>>>> +
>>>>> + dev_dbg(&priv->client->dev, "writing register 0x%02X=0x%02X", reg, val);
>>>>> +
>>>>> + ret = i2c_smbus_write_byte_data(priv->client, reg, val);
>>>>> + if (ret) {
>>>>> + dev_err(&priv->client->dev,
>>>>> + "register write to 0x%02X failed (error %d)",
>>>>> + reg, ret);
>>>>> + }
>>>>
>>>> In case somebody use this driver as heartbeat and writing fails permanently the
>>>> log will be flooded.
>>>
>>> Unless I'm mistaken that would require the device/bus to fail after
>>> successfully probing (probe code itself bails on the first write
>>> failure, so there would be no flooding as a result of that). So while
>>> not impossible, I imagine it would be unlikely, and I'd hate to remove
>>> an error message for such an important condition.
>>>
>>> I suppose I could use dev_err_ratelimited() to soften any potential
>>> flooding, but I second guess that because:
>>>    - In led_core.c set_brightness_delayed() has a dev_err() that would come
>>>      out on each failed LED update anyways.
>>>    - There is precedent in other led drivers of a similar error message.
>>>    - Some userspace logging programs will compresses repeated messages anyways.
>>>
>>> Jacek, what is your preference on this?
>>
>> Let's leave it as is. Permanent I2C bus failure is a critical error and
>> flooding the log would only allow to diagnose the problem quicker.
>>
>>>>> + return ret;
>>>>> +}
>>>>> +
>>>>> +/*
>>>>> + * Custom reset function for IS31FL3216 because it does not have a RESET
>>>>> + * register the way that the other IS31FL32xx chips do. We don't bother
>>>>> + * writing the GPIO and animation registers, because the registers we
>>>>> + * do write ensure those will have no effect.
>>>>> + */
>>>>> +static int is31fl3216_reset(struct is31fl32xx_priv *priv)
>>>>> +{
>>>>> + unsigned int i;
>>>>> + int ret;
>>>>> +
>>>>> + ret = is31fl32xx_write(priv, IS31FL3216_CONFIG_REG,
>>>>> + IS31FL3216_CONFIG_SSD_ENABLE);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> + for (i = 0; i < priv->cdef->channels; i++) {
>>>>> + ret = is31fl32xx_write(priv, priv->cdef->pwm_register_base+i,
>>>>> + 0x00);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> + }
>>>>> + ret = is31fl32xx_write(priv, priv->cdef->pwm_update_reg, 0);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> + ret = is31fl32xx_write(priv, IS31FL3216_LIGHTING_EFFECT_REG, 0x00);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> + ret = is31fl32xx_write(priv, IS31FL3216_CHANNEL_CONFIG_REG, 0x00);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +/*
>>>>> + * Custom Software-Shutdown function for IS31FL3216 because it does not have
>>>>> + * a SHUTDOWN register the way that the other IS31FL32xx chips do.
>>>>> + * We don't bother doing a read/modify/write on the CONFIG register because
>>>>> + * we only ever use a value of '0' for the other fields in that register.
>>>>> + */
>>>>> +static int is31fl3216_software_shutdown(struct is31fl32xx_priv *priv,
>>>>> + bool enable)
>>>>> +{
>>>>> + u8 value = enable ? IS31FL3216_CONFIG_SSD_ENABLE :
>>>>> + IS31FL3216_CONFIG_SSD_DISABLE;
>>>>> +
>>>>> + return is31fl32xx_write(priv, IS31FL3216_CONFIG_REG, value);
>>>>> +}
>>>>> +
>>>>> +/*
>>>>> + * NOTE: A mutex is not needed in this function because:
>>>>> + * - All referenced data is read-only after probe()
>>>>> + * - The I2C core has a mutex on to protect the bus
>>>>> + * - There are no read/modify/write operations
>>>>> + * - Intervening operations between the write of the PWM register
>>>>> + * and the Update register are harmless.
>>>>> + *
>>>>> + * Example:
>>>>> + * PWM_REG_1 write 16
>>>>> + * UPDATE_REG write 0
>>>>> + * PWM_REG_2 write 128
>>>>> + * UPDATE_REG write 0
>>>>> + * vs:
>>>>> + * PWM_REG_1 write 16
>>>>> + * PWM_REG_2 write 128
>>>>> + * UPDATE_REG write 0
>>>>> + * UPDATE_REG write 0
>>>>> + * are equivalent. Poking the Update register merely applies all PWM
>>>>> + * register writes up to that point.
>>>>> + */
>>>>> +static int is31fl32xx_brightness_set(struct led_classdev *led_cdev,
>>>>> + enum led_brightness brightness)
>>>>> +{
>>>>> + const struct is31fl32xx_led_data *led_data =
>>>>> + container_of(led_cdev, struct is31fl32xx_led_data, cdev);
>>>>> + const struct is31fl32xx_chipdef *cdef = led_data->priv->cdef;
>>>>> + u8 pwm_register_offset;
>>>>> + int ret;
>>>>> +
>>>>> + dev_dbg(led_cdev->dev, "%s: %d\n", __func__, brightness);
>>>>> +
>>>>> + /* NOTE: led_data->channel is 1-based */
>>>>> + if (cdef->pwm_registers_reversed)
>>>>> + pwm_register_offset = cdef->channels - led_data->channel;
>>>>> + else
>>>>> + pwm_register_offset = led_data->channel - 1;
>>>>> +
>>>>> + ret = is31fl32xx_write(led_data->priv,
>>>>> + cdef->pwm_register_base + pwm_register_offset,
>>>>> + brightness);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> +
>>>>> + return is31fl32xx_write(led_data->priv, cdef->pwm_update_reg, 0);
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_reset_regs(struct is31fl32xx_priv *priv)
>>>>> +{
>>>>> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>> + int ret;
>>>>> +
>>>>> + if (cdef->reset_reg != IS31FL32XX_REG_NONE) {
>>>>> + ret = is31fl32xx_write(priv, cdef->reset_reg, 0);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> + }
>>>>> +
>>>>> + if (cdef->reset_func)
>>>>> + return cdef->reset_func(priv);
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_software_shutdown(struct is31fl32xx_priv *priv,
>>>>> + bool enable)
>>>>> +{
>>>>> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>> + int ret;
>>>>> +
>>>>> + if (cdef->shutdown_reg != IS31FL32XX_REG_NONE) {
>>>>> + u8 value = enable ? IS31FL32XX_SHUTDOWN_SSD_ENABLE :
>>>>> + IS31FL32XX_SHUTDOWN_SSD_DISABLE;
>>>>> + ret = is31fl32xx_write(priv, cdef->shutdown_reg, value);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> + }
>>>>> +
>>>>> + if (cdef->sw_shutdown_func)
>>>>> + return cdef->sw_shutdown_func(priv, enable);
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_init_regs(struct is31fl32xx_priv *priv)
>>>>> +{
>>>>> + const struct is31fl32xx_chipdef *cdef = priv->cdef;
>>>>> + int ret;
>>>>> +
>>>>> + ret = is31fl32xx_reset_regs(priv);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> +
>>>>> + /*
>>>>> + * Set enable bit for all channels.
>>>>> + * We will control state with PWM registers alone.
>>>>> + */
>>>>> + if (cdef->led_control_register_base != IS31FL32XX_REG_NONE) {
>>>>> + u8 value =
>>>>> + GENMASK(cdef->enable_bits_per_led_control_register-1, 0);
>>>>> + u8 num_regs = cdef->channels /
>>>>> + cdef->enable_bits_per_led_control_register;
>>>>> + int i;
>>>>> +
>>>>> + for (i = 0; i < num_regs; i++) {
>>>>> + ret = is31fl32xx_write(priv,
>>>>> + cdef->led_control_register_base+i,
>>>>> + value);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> + }
>>>>> + }
>>>>> +
>>>>> + ret = is31fl32xx_software_shutdown(priv, false);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> +
>>>>> + if (cdef->global_control_reg != IS31FL32XX_REG_NONE) {
>>>>> + ret = is31fl32xx_write(priv, cdef->global_control_reg, 0x00);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> + }
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +static inline size_t sizeof_is31fl32xx_priv(int num_leds)
>>>>> +{
>>>>> + return sizeof(struct is31fl32xx_priv) +
>>>>> + (sizeof(struct is31fl32xx_led_data) * num_leds);
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_parse_child_dt(const struct device *dev,
>>>>> + const struct device_node *child,
>>>>> + struct is31fl32xx_led_data *led_data)
>>>>> +{
>>>>> + struct led_classdev *cdev = &led_data->cdev;
>>>>> + int ret = 0;
>>>>> + u32 reg;
>>>>> +
>>>>> + if (of_property_read_string(child, "label", &cdev->name))
>>>>> + cdev->name = child->name;
>>>>> +
>>>>> + ret = of_property_read_u32(child, "reg", &reg);
>>>>> + if (ret || reg < 1 || reg > led_data->priv->cdef->channels) {
>>>>> + dev_err(dev,
>>>>> + "Child node %s does not have a valid reg property\n",
>>>>> + child->full_name);
>>>>> + return -EINVAL;
>>>>> + }
>>>>> + led_data->channel = reg;
>>>>> +
>>>>> + of_property_read_string(child, "linux,default-trigger",
>>>>> + &cdev->default_trigger);
>>>>> +
>>>>> + cdev->brightness_set_blocking = is31fl32xx_brightness_set;
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +static struct is31fl32xx_led_data *is31fl32xx_find_led_data(
>>>>> + struct is31fl32xx_priv *priv,
>>>>> + u8 channel)
>>>>> +{
>>>>> + size_t i;
>>>>> +
>>>>> + for (i = 0; i < priv->num_leds; i++) {
>>>>> + if (priv->leds[i].channel == channel)
>>>>> + return &priv->leds[i];
>>>>> + }
>>>>> +
>>>>> + return NULL;
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_parse_dt(struct device *dev,
>>>>> + struct is31fl32xx_priv *priv)
>>>>> +{
>>>>> + struct device_node *child;
>>>>> + int ret = 0;
>>>>> +
>>>>> + for_each_child_of_node(dev->of_node, child) {
>>>>> + struct is31fl32xx_led_data *led_data =
>>>>> + &priv->leds[priv->num_leds];
>>>>
>>>> Maybe i missed something, but is it really protected against out of index
>>>> access?
>>>
>>> The array is allocated with size equal to the number of child nodes,
>>> and num_leds is incremented once for each child node parsed. So in
>>> order for the index to be out of bounds, the number of child nodes
>>> would need to increase during the probe. I assumed that the DT is
>>> static during probing, but if that's not the case then you're right
>>> that this is a potential problem. Also, this equivalent logic is
>>> used in leds-pwm, leds-gpio, and leds-ns2, so that gives me
>>> confidence that its safe.
>>> Unless DT overlays change that assumption?
>>
>> DT overlays would matter here if child DT nodes could be dynamically
>> removed, i.e. if it was possible by design to dynamically unplug LEDs
>> from the current outputs during LED controller operation, which is not
>> the case for this device (and any other LED controller I am aware of).
>>
>>>>> + const struct is31fl32xx_led_data *other_led_data;
>>>>> +
>>>>> + led_data->priv = priv;
>>>>> +
>>>>> + ret = is31fl32xx_parse_child_dt(dev, child, led_data);
>>>>> + if (ret)
>>>>> + goto err;
>>>>> +
>>>>> + /* Detect if channel is already in use by another child */
>>>>> + other_led_data = is31fl32xx_find_led_data(priv,
>>>>> + led_data->channel);
>>>>> + if (other_led_data) {
>>>>> + dev_err(dev,
>>>>> + "%s and %s both attempting to use channel %d\n",
>>>>> + led_data->cdev.name,
>>>>> + other_led_data->cdev.name,
>>>>> + led_data->channel);
>>>>> + goto err;
>>>>> + }
>>>>> +
>>>>> + ret = devm_led_classdev_register(dev, &led_data->cdev);
>>>>> + if (ret) {
>>>>> + dev_err(dev, "failed to register PWM led for %s: %d\n",
>>>>> + led_data->cdev.name, ret);
>>>>> + goto err;
>>>>> + }
>>>>> +
>>>>> + priv->num_leds++;
>>>>> + }
>>>>> +
>>>>> + return 0;
>>>>> +
>>>>> +err:
>>>>> + of_node_put(child);
>>>>> + return ret;
>>>>> +}
>>>>> +
>>>>> +static const struct of_device_id of_is31fl31xx_match[] = {
>>>>> + { .compatible = "issi,is31fl3236", .data = &is31fl3236_cdef, },
>>>>> + { .compatible = "issi,is31fl3235", .data = &is31fl3235_cdef, },
>>>>> + { .compatible = "issi,is31fl3218", .data = &is31fl3218_cdef, },
>>>>> + { .compatible = "issi,is31fl3216", .data = &is31fl3216_cdef, },
>>>>> + {},
>>>>> +};
>>>>> +
>>>>> +MODULE_DEVICE_TABLE(of, of_is31fl31xx_match);
>>>>> +
>>>>> +static int is31fl32xx_probe(struct i2c_client *client,
>>>>> + const struct i2c_device_id *id)
>>>>> +{
>>>>> + const struct is31fl32xx_chipdef *cdef;
>>>>> + const struct of_device_id *of_dev_id;
>>>>> + struct device *dev = &client->dev;
>>>>> + struct is31fl32xx_priv *priv;
>>>>> + int count;
>>>>> + int ret = 0;
>>>>> +
>>>>> + of_dev_id = of_match_device(of_is31fl31xx_match, dev);
>>>>> + if (!of_dev_id)
>>>>> + return -EINVAL;
>>>>> +
>>>>> + cdef = of_dev_id->data;
>>>>> +
>>>>> + count = of_get_child_count(dev->of_node);
>>>>> + if (!count)
>>>>> + return -EINVAL;
>>>>> +
>>>>> + priv = devm_kzalloc(dev, sizeof_is31fl32xx_priv(count),
>>>>> + GFP_KERNEL);
>>>>> + if (!priv)
>>>>> + return -ENOMEM;
>>>>> +
>>>>> + priv->client = client;
>>>>> + priv->cdef = cdef;
>>>>> + i2c_set_clientdata(client, priv);
>>>>> +
>>>>> + ret = is31fl32xx_init_regs(priv);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> +
>>>>> + ret = is31fl32xx_parse_dt(dev, priv);
>>>>> + if (ret)
>>>>> + return ret;
>>>>> +
>>>>> + return 0;
>>>>> +}
>>>>> +
>>>>> +static int is31fl32xx_remove(struct i2c_client *client)
>>>>> +{
>>>>> + struct is31fl32xx_priv *priv = i2c_get_clientdata(client);
>>>>> +
>>>>> + return is31fl32xx_reset_regs(priv);
>>>>> +}
>>>>> +
>>>>> +/*
>>>>> + * i2c-core requires that id_table be non-NULL, even though
>>>>> + * it is not used for DeviceTree based instantiation.
>>>>> + */
>>>>> +static const struct i2c_device_id is31fl31xx_id[] = {
>>>>> + {},
>>>>> +};
>>>>> +
>>>>> +MODULE_DEVICE_TABLE(i2c, is31fl31xx_id);
>>>>> +
>>>>> +static struct i2c_driver is31fl32xx_driver = {
>>>>> + .driver = {
>>>>> + .name = "is31fl32xx",
>>>>> + .of_match_table = of_is31fl31xx_match,
>>>>> + },
>>>>> + .probe = is31fl32xx_probe,
>>>>> + .remove = is31fl32xx_remove,
>>>>
>>>> Sorry, what was the reason to skip shutdown?
>>>
>>> If I understood Jacek's last email on the topic [1] correctly, he's now
>>> of the opinion that the decision to turn LEDs off on reboot should be
>>> left to userspace, rather than done by the driver. For these devices,
>>> the only thing a shutdown callback would do is turn off the LEDs (through
>>> any of multiple methods). So, if we want to leave the state as-is on
>>> reboot there's no need for a shutdown callback.
>>>
>>> [1] http://www.spinics.net/lists/linux-leds/msg05644.html
>>>
>>>>> + .id_table = is31fl31xx_id,
>>>>> +};
>>>>> +
>>>>> +module_i2c_driver(is31fl32xx_driver);
>>>>> +
>>>>> +MODULE_AUTHOR("David Rivshin <drivshin@allworx.com>");
>>>>> +MODULE_DESCRIPTION("ISSI IS31FL32xx LED driver");
>>>>> +MODULE_LICENSE("GPL v2");
>>>>> --
>>>>> 2.5.0
>>>>>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-leds" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

-- 
Best Regards,
Jacek Anaszewski