Skip to contentSkip to conversation

Support for tegra30 thermal.

RFC PATCH New Wei Ni 6 participants 45 messages Feb 18, 2013 – Feb 20, 2013

Subsystems (6)
lore
Share & tools
lore

Wei Ni

author
#0 lore raw

cover Support for tegra30 thermal.

This patch series is to add tegra30 thermal driver.
It use new linux thermal framework, which is developing by Durgadoss.

Note:
1. my series are based on the patches:
http://thread.gmane.org/gmane.linux.power-management.general/30692 , which come from Durgadoss.
And the tree is Rui's -next tree.

Next steps:
1. enhance the tegra30 thermal driver, something like handling the irq, setting the sensor limit value.
2. seprate codes to create a new files which is for cooling device, something like tegra3_cooling.c. it will handle dvfs.

Wei Ni (9):
ARM: dt: t30 cardhu: add dt entry for lm90
hwmon: (lm90) split set&show temp as common codes
hwmon: (lm90) add support to handle irq
hwmon: (lm90) use macros for the indexes of temp8 and temp11
Thermal: Support using dt node to get sensor
hwmon: (lm90) Register to the thermal framework
thermal: tegra30: add tegra30 thermal driver
ARM: dt: t30 cardhu: add dt entry for thermal driver
ARM: tegra: defconfig: enable thermal framework

 .../devicetree/bindings/thermal/tegra3-thermal.txt |   41 ++
 arch/arm/boot/dts/tegra30-cardhu.dtsi              |   27 +
 arch/arm/configs/tegra_defconfig                   |    2 +
 drivers/hwmon/lm90.c                               |  514 +++++++++++++++-----
 drivers/thermal/Kconfig                            |    9 +
 drivers/thermal/Makefile                           |    1 +
 drivers/thermal/tegra3_thermal.c                   |  289 +++++++++++
 drivers/thermal/thermal_sys.c                      |   30 ++
 include/linux/thermal.h                            |    9 +
 9 files changed, 812 insertions(+), 110 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/thermal/tegra3-thermal.txt
 create mode 100644 drivers/thermal/tegra3_thermal.c
Signature · 1 line
1.7.9.5
author
#1 lore raw

RFC PATCH 1/9 ARM: dt: t30 cardhu: add dt entry for lm90

Enable thermal sensor lm90 in the dts file.

---
 arch/arm/boot/dts/tegra30-cardhu.dtsi |    7 +++++++
 1 file changed, 7 insertions(+)
arch/arm/boot/dts/tegra30-cardhu.dtsi +7 -0
index bdb2a66..15ad1ad 100644
@@ -273,6 +273,13 @@
 				};
 			};
 		};
+
+		nct1008: lm90 at 4c {
+			compatible = "lm90";
+			reg = <0x4c>;
+			interrupt-parent = <&gpio>;
+			interrupts = <226 0x08>; /* gpio PCC2 */
+		};
 	};
 
 	spi at 7000da00 {
Signature · 1 line
1.7.9.5
#2 lore raw

RFC PATCH 1/9 ARM: dt: t30 cardhu: add dt entry for lm90

Quoted 1 line · Wei Ni

Enable thermal sensor lm90 in the dts file.

Btw, shouldn't this patch come last, after all the changes you did to lm90? If you keep the current order, you will need to ensure that the device performs as expected on Cardhu for each of the remaining patches of the series. It is probably safer to just move this one at the end to avoid trouble.

Alex.

author
#3 lore raw

RFC PATCH 1/9 ARM: dt: t30 cardhu: add dt entry for lm90

Quoted 10 lines · Alex Courbot
Quoted 1 line · Wei Ni

Enable thermal sensor lm90 in the dts file.

Btw, shouldn't this patch come last, after all the changes you did to lm90? If you keep the current order, you will need to ensure that the device performs as expected on Cardhu for each of the remaining patches of the series. It is probably safer to just move this one at the end to avoid trouble.

I will try split this patch set into 3 separate sets in my next version, like Matthew's suggestion, so the patches will be clear.

Wei.

Quoted 3 lines

Alex.

#4 lore raw

RFC PATCH 1/9 ARM: dt: t30 cardhu: add dt entry for lm90

In reply to Alex Courbot
Quoted 10 lines · Alex Courbot
Quoted 1 line · Wei Ni

Enable thermal sensor lm90 in the dts file.

Btw, shouldn't this patch come last, after all the changes you did to lm90? If you keep the current order, you will need to ensure that the device performs as expected on Cardhu for each of the remaining patches of the series. It is probably safer to just move this one at the end to avoid trouble.

Hmm. This patch on its own seems simple enough, and so I should really just apply it as soon as 3.9-rc1 is out.

Note that we (and probably others) have many device trees for boards besides Cardhu that instantiate the LM90 driver. The driver has to continue to work perfectly for all the patches in the series. Not enabling the driver on Cardhu isn't enough to allow patches in the middle of the series to break the already-working driver.

author
#5 lore raw

RFC PATCH 2/9 hwmon: (lm90) split set&show temp as common codes

In reply to Wei Ni

Split set&show temp codes as common functions, so we can use it directly when implement linux thermal framework.

---
 drivers/hwmon/lm90.c |  117 +++++++++++++++++++++++++++++++-------------------
 1 file changed, 72 insertions(+), 45 deletions(-)
drivers/hwmon/lm90.c +72 -45
index 863412a..caf01b0 100644
@@ -702,29 +702,35 @@ static u16 temp_to_u16_adt7461(struct lm90_data *data, long val)
  * Sysfs stuff
  */
 
-static ssize_t show_temp8(struct device *dev, struct device_attribute *devattr,
-			  char *buf)
+static void _show_temp8(struct device *dev, int index, int *temp)
 {
-	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
 	struct lm90_data *data = lm90_update_device(dev);
-	int temp;
 
 	if (data->kind == adt7461)
-		temp = temp_from_u8_adt7461(data, data->temp8[attr->index]);
+		*temp = temp_from_u8_adt7461(data, data->temp8[index]);
 	else if (data->kind == max6646)
-		temp = temp_from_u8(data->temp8[attr->index]);
+		*temp = temp_from_u8(data->temp8[index]);
 	else
-		temp = temp_from_s8(data->temp8[attr->index]);
+		*temp = temp_from_s8(data->temp8[index]);
 
 	/* +16 degrees offset for temp2 for the LM99 */
-	if (data->kind == lm99 && attr->index == 3)
-		temp += 16000;
+	if (data->kind == lm99 && index == 3)
+		*temp += 16000;
+
+}
+
+static ssize_t show_temp8(struct device *dev, struct device_attribute *devattr,
+			  char *buf)
+{
+	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
+	int temp;
+
+	_show_temp8(dev, attr->index, &temp);
 
 	return sprintf(buf, "%d\n", temp);
 }
 
-static ssize_t set_temp8(struct device *dev, struct device_attribute *devattr,
-			 const char *buf, size_t count)
+static void _set_temp8(struct device *dev, int index, long val)
 {
 	static const u8 reg[8] = {
 		LM90_REG_W_LOCAL_LOW,
@@ -737,60 +743,73 @@ static ssize_t set_temp8(struct device *dev, struct device_attribute *devattr,
 		MAX6659_REG_W_REMOTE_EMERG,
 	};
 
-	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
 	struct i2c_client *client = to_i2c_client(dev);
 	struct lm90_data *data = i2c_get_clientdata(client);
-	int nr = attr->index;
-	long val;
-	int err;
-
-	err = kstrtol(buf, 10, &val);
-	if (err < 0)
-		return err;
 
 	/* +16 degrees offset for temp2 for the LM99 */
-	if (data->kind == lm99 && attr->index == 3)
+	if (data->kind == lm99 && index == 3)
 		val -= 16000;
 
 	mutex_lock(&data->update_lock);
 	if (data->kind == adt7461)
-		data->temp8[nr] = temp_to_u8_adt7461(data, val);
+		data->temp8[index] = temp_to_u8_adt7461(data, val);
 	else if (data->kind == max6646)
-		data->temp8[nr] = temp_to_u8(val);
+		data->temp8[index] = temp_to_u8(val);
 	else
-		data->temp8[nr] = temp_to_s8(val);
+		data->temp8[index] = temp_to_s8(val);
 
-	lm90_select_remote_channel(client, data, nr >= 6);
-	i2c_smbus_write_byte_data(client, reg[nr], data->temp8[nr]);
+	lm90_select_remote_channel(client, data, index >= 6);
+	i2c_smbus_write_byte_data(client, reg[index], data->temp8[index]);
 	lm90_select_remote_channel(client, data, 0);
 
 	mutex_unlock(&data->update_lock);
+}
+
+static ssize_t set_temp8(struct device *dev, struct device_attribute *devattr,
+			 const char *buf, size_t count)
+{
+	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
+	int index = attr->index;
+	long val;
+	int err;
+
+	err = kstrtol(buf, 10, &val);
+	if (err < 0)
+		return err;
+
+	_set_temp8(dev, index, val);
+
 	return count;
 }
 
-static ssize_t show_temp11(struct device *dev, struct device_attribute *devattr,
-			   char *buf)
+static void _show_temp11(struct device *dev, int index, int *temp)
 {
-	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
 	struct lm90_data *data = lm90_update_device(dev);
-	int temp;
 
 	if (data->kind == adt7461)
-		temp = temp_from_u16_adt7461(data, data->temp11[attr->index]);
+		*temp = temp_from_u16_adt7461(data, data->temp11[index]);
 	else if (data->kind == max6646)
-		temp = temp_from_u16(data->temp11[attr->index]);
+		*temp = temp_from_u16(data->temp11[index]);
 	else
-		temp = temp_from_s16(data->temp11[attr->index]);
+		*temp = temp_from_s16(data->temp11[index]);
 
 	/* +16 degrees offset for temp2 for the LM99 */
-	if (data->kind == lm99 &&  attr->index <= 2)
-		temp += 16000;
+	if (data->kind == lm99 &&  index <= 2)
+		*temp += 16000;
+}
+
+static ssize_t show_temp11(struct device *dev, struct device_attribute *devattr,
+			   char *buf)
+{
+	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
+	int temp;
+
+	_show_temp11(dev, attr->index, &temp);
 
 	return sprintf(buf, "%d\n", temp);
 }
 
-static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
-			  const char *buf, size_t count)
+static void _set_temp11(struct device *dev, int nr, int index, long val)
 {
 	struct {
 		u8 high;
@@ -804,17 +823,8 @@ static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
 		{ LM90_REG_W_REMOTE_HIGHH, LM90_REG_W_REMOTE_HIGHL, 1 }
 	};
 
-	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
 	struct i2c_client *client = to_i2c_client(dev);
 	struct lm90_data *data = i2c_get_clientdata(client);
-	int nr = attr->nr;
-	int index = attr->index;
-	long val;
-	int err;
-
-	err = kstrtol(buf, 10, &val);
-	if (err < 0)
-		return err;
 
 	/* +16 degrees offset for temp2 for the LM99 */
 	if (data->kind == lm99 && index <= 2)
@@ -839,6 +849,23 @@ static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
 	lm90_select_remote_channel(client, data, 0);
 
 	mutex_unlock(&data->update_lock);
+}
+
+static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
+			  const char *buf, size_t count)
+{
+	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
+	int nr = attr->nr;
+	int index = attr->index;
+	long val;
+	int err;
+
+	err = kstrtol(buf, 10, &val);
+	if (err < 0)
+		return err;
+
+	_set_temp11(dev, nr, index, val);
+
 	return count;
 }
 
Signature · 1 line
1.7.9.5
#6 lore raw

RFC PATCH 2/9 hwmon: (lm90) split set&show temp as common codes

Wei,

This set of patches seems to include 3 logical subsets
1) lm90 changes
2) dt support in drivers/thermal
3) tegra30 support in drivers/thermal

Would it make sense to split this into 3 separate sets of changes?

-Matt Longnecker

nvpublic

author
#7 lore raw

RFC PATCH 2/9 hwmon: (lm90) split set&show temp as common codes

Quoted 8 lines · Matthew Longnecker

Wei,

This set of patches seems to include 3 logical subsets
1) lm90 changes
2) dt support in drivers/thermal
3) tegra30 support in drivers/thermal

Would it make sense to split this into 3 separate sets of changes?

Ok, I will do it in my next version.

Quoted 5 lines

-Matt Longnecker

nvpublic

maintainer
#8 lore raw

RFC PATCH 2/9 hwmon: (lm90) split set&show temp as common codes

In reply to Wei Ni
Quoted 13 lines · Wei Ni

Split set&show temp codes as common functions, so we can use it directly when implement linux thermal framework.

---
 drivers/hwmon/lm90.c |  117 +++++++++++++++++++++++++++++++-------------------
 1 file changed, 72 insertions(+), 45 deletions(-)
drivers/hwmon/lm90.c +0 -0
index 863412a..caf01b0 100644

drivers/hwmon/lm90.c static u16 temp_to_u16_adt7461(struct lm90_data *data, long val) · line 705

Quoted 6 lines
drivers/hwmon/lm90.c +1 -2
@@ -702,29 +702,35 @@ static u16 temp_to_u16_adt7461(struct lm90_data *data, long val)
  * Sysfs stuff
  */
 
-static ssize_t show_temp8(struct device *dev, struct device_attribute *devattr,
-			  char *buf)
+static void _show_temp8(struct device *dev, int index, int *temp)

Why do the new functions not simply return the value ?
Void return and returning the value in a pointer doesn't make much sense to me.

Guenter

Quoted 190 lines
drivers/hwmon/lm90.c +71 -43
 {
-	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
 	struct lm90_data *data = lm90_update_device(dev);
-	int temp;
 
 	if (data->kind == adt7461)
-		temp = temp_from_u8_adt7461(data, data->temp8[attr->index]);
+		*temp = temp_from_u8_adt7461(data, data->temp8[index]);
 	else if (data->kind == max6646)
-		temp = temp_from_u8(data->temp8[attr->index]);
+		*temp = temp_from_u8(data->temp8[index]);
 	else
-		temp = temp_from_s8(data->temp8[attr->index]);
+		*temp = temp_from_s8(data->temp8[index]);
 
 	/* +16 degrees offset for temp2 for the LM99 */
-	if (data->kind == lm99 && attr->index == 3)
-		temp += 16000;
+	if (data->kind == lm99 && index == 3)
+		*temp += 16000;
+
+}
+
+static ssize_t show_temp8(struct device *dev, struct device_attribute *devattr,
+			  char *buf)
+{
+	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
+	int temp;
+
+	_show_temp8(dev, attr->index, &temp);
 
 	return sprintf(buf, "%d\n", temp);
 }
 
-static ssize_t set_temp8(struct device *dev, struct device_attribute *devattr,
-			 const char *buf, size_t count)
+static void _set_temp8(struct device *dev, int index, long val)
 {
 	static const u8 reg[8] = {
 		LM90_REG_W_LOCAL_LOW,
@@ -737,60 +743,73 @@ static ssize_t set_temp8(struct device *dev, struct device_attribute *devattr,
 		MAX6659_REG_W_REMOTE_EMERG,
 	};
 
-	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
 	struct i2c_client *client = to_i2c_client(dev);
 	struct lm90_data *data = i2c_get_clientdata(client);
-	int nr = attr->index;
-	long val;
-	int err;
-
-	err = kstrtol(buf, 10, &val);
-	if (err < 0)
-		return err;
 
 	/* +16 degrees offset for temp2 for the LM99 */
-	if (data->kind == lm99 && attr->index == 3)
+	if (data->kind == lm99 && index == 3)
 		val -= 16000;
 
 	mutex_lock(&data->update_lock);
 	if (data->kind == adt7461)
-		data->temp8[nr] = temp_to_u8_adt7461(data, val);
+		data->temp8[index] = temp_to_u8_adt7461(data, val);
 	else if (data->kind == max6646)
-		data->temp8[nr] = temp_to_u8(val);
+		data->temp8[index] = temp_to_u8(val);
 	else
-		data->temp8[nr] = temp_to_s8(val);
+		data->temp8[index] = temp_to_s8(val);
 
-	lm90_select_remote_channel(client, data, nr >= 6);
-	i2c_smbus_write_byte_data(client, reg[nr], data->temp8[nr]);
+	lm90_select_remote_channel(client, data, index >= 6);
+	i2c_smbus_write_byte_data(client, reg[index], data->temp8[index]);
 	lm90_select_remote_channel(client, data, 0);
 
 	mutex_unlock(&data->update_lock);
+}
+
+static ssize_t set_temp8(struct device *dev, struct device_attribute *devattr,
+			 const char *buf, size_t count)
+{
+	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
+	int index = attr->index;
+	long val;
+	int err;
+
+	err = kstrtol(buf, 10, &val);
+	if (err < 0)
+		return err;
+
+	_set_temp8(dev, index, val);
+
 	return count;
 }
 
-static ssize_t show_temp11(struct device *dev, struct device_attribute *devattr,
-			   char *buf)
+static void _show_temp11(struct device *dev, int index, int *temp)
 {
-	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
 	struct lm90_data *data = lm90_update_device(dev);
-	int temp;
 
 	if (data->kind == adt7461)
-		temp = temp_from_u16_adt7461(data, data->temp11[attr->index]);
+		*temp = temp_from_u16_adt7461(data, data->temp11[index]);
 	else if (data->kind == max6646)
-		temp = temp_from_u16(data->temp11[attr->index]);
+		*temp = temp_from_u16(data->temp11[index]);
 	else
-		temp = temp_from_s16(data->temp11[attr->index]);
+		*temp = temp_from_s16(data->temp11[index]);
 
 	/* +16 degrees offset for temp2 for the LM99 */
-	if (data->kind == lm99 &&  attr->index <= 2)
-		temp += 16000;
+	if (data->kind == lm99 &&  index <= 2)
+		*temp += 16000;
+}
+
+static ssize_t show_temp11(struct device *dev, struct device_attribute *devattr,
+			   char *buf)
+{
+	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
+	int temp;
+
+	_show_temp11(dev, attr->index, &temp);
 
 	return sprintf(buf, "%d\n", temp);
 }
 
-static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
-			  const char *buf, size_t count)
+static void _set_temp11(struct device *dev, int nr, int index, long val)
 {
 	struct {
 		u8 high;
@@ -804,17 +823,8 @@ static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
 		{ LM90_REG_W_REMOTE_HIGHH, LM90_REG_W_REMOTE_HIGHL, 1 }
 	};
 
-	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
 	struct i2c_client *client = to_i2c_client(dev);
 	struct lm90_data *data = i2c_get_clientdata(client);
-	int nr = attr->nr;
-	int index = attr->index;
-	long val;
-	int err;
-
-	err = kstrtol(buf, 10, &val);
-	if (err < 0)
-		return err;
 
 	/* +16 degrees offset for temp2 for the LM99 */
 	if (data->kind == lm99 && index <= 2)
@@ -839,6 +849,23 @@ static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
 	lm90_select_remote_channel(client, data, 0);
 
 	mutex_unlock(&data->update_lock);
+}
+
+static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
+			  const char *buf, size_t count)
+{
+	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
+	int nr = attr->nr;
+	int index = attr->index;
+	long val;
+	int err;
+
+	err = kstrtol(buf, 10, &val);
+	if (err < 0)
+		return err;
+
+	_set_temp11(dev, nr, index, val);
+
 	return count;
 }
 
Signature · 7 lines
1.7.9.5


_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
author
#9 lore raw

RFC PATCH 2/9 hwmon: (lm90) split set&show temp as common codes

Quoted 17 lines · Guenter Roeck
Quoted 16 lines · Wei Ni

Split set&show temp codes as common functions, so we can use it directly when implement linux thermal framework.

---
 drivers/hwmon/lm90.c |  117 +++++++++++++++++++++++++++++++-------------------
 1 file changed, 72 insertions(+), 45 deletions(-)
drivers/hwmon/lm90.c +0 -0
index 863412a..caf01b0 100644
@@ -702,29 +702,35 @@ static u16 temp_to_u16_adt7461(struct lm90_data *data, long val)
  * Sysfs stuff
  */
 
Quoted 6 lines

drivers/hwmon/lm90.c static u16 temp_to_u16_adt7461(struct lm90_data *data, long val) · line 705

Quoted 3 lines
drivers/hwmon/lm90.c +1 -2
-static ssize_t show_temp8(struct device *dev, struct device_attribute *devattr,
-			  char *buf)
+static void _show_temp8(struct device *dev, int index, int *temp)

Why do the new functions not simply return the value ?
Void return and returning the value in a pointer doesn't make much sense to me.

Yes, you are right, it's better to return value simply. I will change it in my next version.

Wei.

Quoted 193 lines

Guenter

Quoted 190 lines
drivers/hwmon/lm90.c +71 -43
 {
-	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
 	struct lm90_data *data = lm90_update_device(dev);
-	int temp;
 
 	if (data->kind == adt7461)
-		temp = temp_from_u8_adt7461(data, data->temp8[attr->index]);
+		*temp = temp_from_u8_adt7461(data, data->temp8[index]);
 	else if (data->kind == max6646)
-		temp = temp_from_u8(data->temp8[attr->index]);
+		*temp = temp_from_u8(data->temp8[index]);
 	else
-		temp = temp_from_s8(data->temp8[attr->index]);
+		*temp = temp_from_s8(data->temp8[index]);
 
 	/* +16 degrees offset for temp2 for the LM99 */
-	if (data->kind == lm99 && attr->index == 3)
-		temp += 16000;
+	if (data->kind == lm99 && index == 3)
+		*temp += 16000;
+
+}
+
+static ssize_t show_temp8(struct device *dev, struct device_attribute *devattr,
+			  char *buf)
+{
+	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
+	int temp;
+
+	_show_temp8(dev, attr->index, &temp);
 
 	return sprintf(buf, "%d\n", temp);
 }
 
-static ssize_t set_temp8(struct device *dev, struct device_attribute *devattr,
-			 const char *buf, size_t count)
+static void _set_temp8(struct device *dev, int index, long val)
 {
 	static const u8 reg[8] = {
 		LM90_REG_W_LOCAL_LOW,
@@ -737,60 +743,73 @@ static ssize_t set_temp8(struct device *dev, struct device_attribute *devattr,
 		MAX6659_REG_W_REMOTE_EMERG,
 	};
 
-	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
 	struct i2c_client *client = to_i2c_client(dev);
 	struct lm90_data *data = i2c_get_clientdata(client);
-	int nr = attr->index;
-	long val;
-	int err;
-
-	err = kstrtol(buf, 10, &val);
-	if (err < 0)
-		return err;
 
 	/* +16 degrees offset for temp2 for the LM99 */
-	if (data->kind == lm99 && attr->index == 3)
+	if (data->kind == lm99 && index == 3)
 		val -= 16000;
 
 	mutex_lock(&data->update_lock);
 	if (data->kind == adt7461)
-		data->temp8[nr] = temp_to_u8_adt7461(data, val);
+		data->temp8[index] = temp_to_u8_adt7461(data, val);
 	else if (data->kind == max6646)
-		data->temp8[nr] = temp_to_u8(val);
+		data->temp8[index] = temp_to_u8(val);
 	else
-		data->temp8[nr] = temp_to_s8(val);
+		data->temp8[index] = temp_to_s8(val);
 
-	lm90_select_remote_channel(client, data, nr >= 6);
-	i2c_smbus_write_byte_data(client, reg[nr], data->temp8[nr]);
+	lm90_select_remote_channel(client, data, index >= 6);
+	i2c_smbus_write_byte_data(client, reg[index], data->temp8[index]);
 	lm90_select_remote_channel(client, data, 0);
 
 	mutex_unlock(&data->update_lock);
+}
+
+static ssize_t set_temp8(struct device *dev, struct device_attribute *devattr,
+			 const char *buf, size_t count)
+{
+	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
+	int index = attr->index;
+	long val;
+	int err;
+
+	err = kstrtol(buf, 10, &val);
+	if (err < 0)
+		return err;
+
+	_set_temp8(dev, index, val);
+
 	return count;
 }
 
-static ssize_t show_temp11(struct device *dev, struct device_attribute *devattr,
-			   char *buf)
+static void _show_temp11(struct device *dev, int index, int *temp)
 {
-	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
 	struct lm90_data *data = lm90_update_device(dev);
-	int temp;
 
 	if (data->kind == adt7461)
-		temp = temp_from_u16_adt7461(data, data->temp11[attr->index]);
+		*temp = temp_from_u16_adt7461(data, data->temp11[index]);
 	else if (data->kind == max6646)
-		temp = temp_from_u16(data->temp11[attr->index]);
+		*temp = temp_from_u16(data->temp11[index]);
 	else
-		temp = temp_from_s16(data->temp11[attr->index]);
+		*temp = temp_from_s16(data->temp11[index]);
 
 	/* +16 degrees offset for temp2 for the LM99 */
-	if (data->kind == lm99 &&  attr->index <= 2)
-		temp += 16000;
+	if (data->kind == lm99 &&  index <= 2)
+		*temp += 16000;
+}
+
+static ssize_t show_temp11(struct device *dev, struct device_attribute *devattr,
+			   char *buf)
+{
+	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
+	int temp;
+
+	_show_temp11(dev, attr->index, &temp);
 
 	return sprintf(buf, "%d\n", temp);
 }
 
-static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
-			  const char *buf, size_t count)
+static void _set_temp11(struct device *dev, int nr, int index, long val)
 {
 	struct {
 		u8 high;
@@ -804,17 +823,8 @@ static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
 		{ LM90_REG_W_REMOTE_HIGHH, LM90_REG_W_REMOTE_HIGHL, 1 }
 	};
 
-	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
 	struct i2c_client *client = to_i2c_client(dev);
 	struct lm90_data *data = i2c_get_clientdata(client);
-	int nr = attr->nr;
-	int index = attr->index;
-	long val;
-	int err;
-
-	err = kstrtol(buf, 10, &val);
-	if (err < 0)
-		return err;
 
 	/* +16 degrees offset for temp2 for the LM99 */
 	if (data->kind == lm99 && index <= 2)
@@ -839,6 +849,23 @@ static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
 	lm90_select_remote_channel(client, data, 0);
 
 	mutex_unlock(&data->update_lock);
+}
+
+static ssize_t set_temp11(struct device *dev, struct device_attribute *devattr,
+			  const char *buf, size_t count)
+{
+	struct sensor_device_attribute_2 *attr = to_sensor_dev_attr_2(devattr);
+	int nr = attr->nr;
+	int index = attr->index;
+	long val;
+	int err;
+
+	err = kstrtol(buf, 10, &val);
+	if (err < 0)
+		return err;
+
+	_set_temp11(dev, nr, index, val);
+
 	return count;
 }
 
Signature · 7 lines
1.7.9.5


_______________________________________________
lm-sensors mailing list
lm-sensors at lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
#10 lore raw

RFC PATCH 2/9 hwmon: (lm90) split set&show temp as common codes

In reply to Wei Ni

drivers/hwmon/lm90.c static u16 temp_to_u16_adt7461(struct lm90_data *data, long val) · line 705

Quoted 8 lines · Wei Ni

Split set&show temp codes as common functions, so we can use it directly when implement linux thermal framework.

drivers/hwmon/lm90.c +1 -2
 
-static ssize_t show_temp8(struct device *dev, struct device_attribute *devattr,
-			  char *buf)
+static void _show_temp8(struct device *dev, int index, int *temp)

This function now seems to be "get" not "show". If you rename it to e.g. "get_temp8", then you won't need to use an _ in the function name. If that name causes conflicts with later patches, perhaps "read_temp8".

drivers/hwmon/lm90.c static u16 temp_to_u16_adt7461(struct lm90_data *data, long val) · line 733

Quoted 3 lines
drivers/hwmon/lm90.c +1 -2
-static ssize_t set_temp8(struct device *dev, struct device_attribute *devattr,
-			 const char *buf, size_t count)
+static void _set_temp8(struct device *dev, int index, long val)

Similarly, perhaps "write_temp8"?

Similar comments for the rest of the patch.

author
#11 lore raw

RFC PATCH 2/9 hwmon: (lm90) split set&show temp as common codes

Quoted 6 lines · Stephen Warren
Quoted 4 lines · Wei Ni

Split set&show temp codes as common functions, so we can use it directly when implement linux thermal framework.

drivers/hwmon/lm90.c +0 -0
Quoted 7 lines

drivers/hwmon/lm90.c static u16 temp_to_u16_adt7461(struct lm90_data *data, long val) · line 705

Quoted 3 lines
drivers/hwmon/lm90.c +1 -2
-static ssize_t show_temp8(struct device *dev, struct device_attribute *devattr,
-			  char *buf)
+static void _show_temp8(struct device *dev, int index, int *temp)

This function now seems to be "get" not "show". If you rename it to e.g. "get_temp8", then you won't need to use an _ in the function name. If that name causes conflicts with later patches, perhaps "read_temp8".

Yes, it looks "read_" is better.
I will change it.

Quoted 9 lines

drivers/hwmon/lm90.c static u16 temp_to_u16_adt7461(struct lm90_data *data, long val) · line 733

Quoted 4 lines
drivers/hwmon/lm90.c +1 -2
-static ssize_t set_temp8(struct device *dev, struct device_attribute *devattr,
-			 const char *buf, size_t count)
+static void _set_temp8(struct device *dev, int index, long val)

Similarly, perhaps "write_temp8"?

Similar comments for the rest of the patch.

author
#12 lore raw

RFC PATCH 3/9 hwmon: (lm90) add support to handle irq

In reply to Wei Ni

Add support to handle irq. When the temperature touch the limit value, the driver can handle the interrupt.

---
 drivers/hwmon/lm90.c |   38 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
drivers/hwmon/lm90.c +38 -0
index caf01b0..80311ef 100644
@@ -89,6 +89,8 @@
 #include <linux/err.h>
 #include <linux/mutex.h>
 #include <linux/sysfs.h>
+#include <linux/interrupt.h>
+#include <linux/of_irq.h>
 
 /*
  * Addresses to scan
@@ -302,6 +304,7 @@ static const struct lm90_params lm90_params[] = {
 struct lm90_data {
 	struct device *hwmon_dev;
 	struct mutex update_lock;
+	struct work_struct irq_work;
 	char valid; /* zero until following fields are valid */
 	unsigned long last_updated; /* in jiffies */
 	int kind;
@@ -1418,6 +1421,29 @@ static void lm90_init_client(struct i2c_client *client)
 		i2c_smbus_write_byte_data(client, LM90_REG_W_CONFIG1, config);
 }
 
+static void lm90_alert(struct i2c_client *client, unsigned int flag);
+
+static void lm90_irq_work(struct work_struct *work)
+{
+	struct lm90_data *data = container_of(work, struct lm90_data,
+					      irq_work);
+	struct i2c_client *client = to_i2c_client(data->hwmon_dev->parent);
+
+	lm90_alert(client, 0);
+
+	enable_irq(client->irq);
+}
+
+static irqreturn_t lm90_irq(int irq, void *dev_id)
+{
+	struct lm90_data *data = dev_id;
+
+	disable_irq_nosync(irq);
+	schedule_work(&data->irq_work);
+
+	return IRQ_HANDLED;
+}
+
 static int lm90_probe(struct i2c_client *client,
 		      const struct i2c_device_id *id)
 {
@@ -1494,6 +1520,17 @@ static int lm90_probe(struct i2c_client *client,
 		goto exit_remove_files;
 	}
 
+	if (client->irq >= 0) {
+		INIT_WORK(&data->irq_work, lm90_irq_work);
+		dev_dbg(dev, "lm90 irq: %d\n", client->irq);
+		err = request_irq(client->irq, lm90_irq, IRQF_TRIGGER_LOW,
+				       "lm90", data);
+		if (err < 0) {
+			dev_err(dev, "cannot request interrupt\n");
+			goto exit_remove_files;
+		}
+	}
+
 	return 0;
 
 exit_remove_files:
@@ -1507,6 +1544,7 @@ static int lm90_remove(struct i2c_client *client)
 {
 	struct lm90_data *data = i2c_get_clientdata(client);
 
+	free_irq(client->irq, data);
 	hwmon_device_unregister(data->hwmon_dev);
 	lm90_remove_files(client, data);
 	lm90_restore_conf(client, data);
Signature · 1 line
1.7.9.5
maintainer
#13 lore raw

RFC PATCH 3/9 hwmon: (lm90) add support to handle irq

Quoted 61 lines · Wei Ni

Add support to handle irq. When the temperature touch the limit value, the driver can handle the interrupt.

---
 drivers/hwmon/lm90.c |   38 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
drivers/hwmon/lm90.c +26 -0
index caf01b0..80311ef 100644
@@ -89,6 +89,8 @@
 #include <linux/err.h>
 #include <linux/mutex.h>
 #include <linux/sysfs.h>
+#include <linux/interrupt.h>
+#include <linux/of_irq.h>
 
 /*
  * Addresses to scan
@@ -302,6 +304,7 @@ static const struct lm90_params lm90_params[] = {
 struct lm90_data {
 	struct device *hwmon_dev;
 	struct mutex update_lock;
+	struct work_struct irq_work;
 	char valid; /* zero until following fields are valid */
 	unsigned long last_updated; /* in jiffies */
 	int kind;
@@ -1418,6 +1421,29 @@ static void lm90_init_client(struct i2c_client *client)
 		i2c_smbus_write_byte_data(client, LM90_REG_W_CONFIG1, config);
 }
 
+static void lm90_alert(struct i2c_client *client, unsigned int flag);
+
+static void lm90_irq_work(struct work_struct *work)
+{
+	struct lm90_data *data = container_of(work, struct lm90_data,
+					      irq_work);
+	struct i2c_client *client = to_i2c_client(data->hwmon_dev->parent);
+
+	lm90_alert(client, 0);
+
+	enable_irq(client->irq);
+}
+
+static irqreturn_t lm90_irq(int irq, void *dev_id)
+{
+	struct lm90_data *data = dev_id;
+
+	disable_irq_nosync(irq);
+	schedule_work(&data->irq_work);
+
+	return IRQ_HANDLED;
+}
+
 static int lm90_probe(struct i2c_client *client,
 		      const struct i2c_device_id *id)
 {

drivers/hwmon/lm90.c static int lm90_probe(struct i2c_client *client, · line 1527

Quoted 8 lines
drivers/hwmon/lm90.c +5 -0
@@ -1494,6 +1520,17 @@ static int lm90_probe(struct i2c_client *client,
 		goto exit_remove_files;
 	}
 
+	if (client->irq >= 0) {
+		INIT_WORK(&data->irq_work, lm90_irq_work);
+		dev_dbg(dev, "lm90 irq: %d\n", client->irq);
+		err = request_irq(client->irq, lm90_irq, IRQF_TRIGGER_LOW,
+				       "lm90", data);

request_threaded_irq or even better devm_request_threaded_irq would be better here.

Quoted 26 lines
drivers/hwmon/lm90.c +7 -0
+		if (err < 0) {
+			dev_err(dev, "cannot request interrupt\n");
+			goto exit_remove_files;
+		}
+	}
+
 	return 0;
 
 exit_remove_files:
@@ -1507,6 +1544,7 @@ static int lm90_remove(struct i2c_client *client)
 {
 	struct lm90_data *data = i2c_get_clientdata(client);
 
+	free_irq(client->irq, data);
 	hwmon_device_unregister(data->hwmon_dev);
 	lm90_remove_files(client, data);
 	lm90_restore_conf(client, data);
Signature · 7 lines
1.7.9.5


_______________________________________________
lm-sensors mailing list
lm-sensors at lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
author
#14 lore raw

RFC PATCH 3/9 hwmon: (lm90) add support to handle irq

Quoted 65 lines · Guenter Roeck
Quoted 64 lines · Wei Ni

Add support to handle irq. When the temperature touch the limit value, the driver can handle the interrupt.

---
 drivers/hwmon/lm90.c |   38 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
drivers/hwmon/lm90.c +26 -0
index caf01b0..80311ef 100644
@@ -89,6 +89,8 @@
 #include <linux/err.h>
 #include <linux/mutex.h>
 #include <linux/sysfs.h>
+#include <linux/interrupt.h>
+#include <linux/of_irq.h>
 
 /*
  * Addresses to scan
@@ -302,6 +304,7 @@ static const struct lm90_params lm90_params[] = {
 struct lm90_data {
 	struct device *hwmon_dev;
 	struct mutex update_lock;
+	struct work_struct irq_work;
 	char valid; /* zero until following fields are valid */
 	unsigned long last_updated; /* in jiffies */
 	int kind;
@@ -1418,6 +1421,29 @@ static void lm90_init_client(struct i2c_client *client)
 		i2c_smbus_write_byte_data(client, LM90_REG_W_CONFIG1, config);
 }
 
+static void lm90_alert(struct i2c_client *client, unsigned int flag);
+
+static void lm90_irq_work(struct work_struct *work)
+{
+	struct lm90_data *data = container_of(work, struct lm90_data,
+					      irq_work);
+	struct i2c_client *client = to_i2c_client(data->hwmon_dev->parent);
+
+	lm90_alert(client, 0);
+
+	enable_irq(client->irq);
+}
+
+static irqreturn_t lm90_irq(int irq, void *dev_id)
+{
+	struct lm90_data *data = dev_id;
+
+	disable_irq_nosync(irq);
+	schedule_work(&data->irq_work);
+
+	return IRQ_HANDLED;
+}
+
 static int lm90_probe(struct i2c_client *client,
 		      const struct i2c_device_id *id)
 {
@@ -1494,6 +1520,17 @@ static int lm90_probe(struct i2c_client *client,
 		goto exit_remove_files;
 	}
 
Quoted 7 lines

drivers/hwmon/lm90.c static int lm90_probe(struct i2c_client *client, · line 1527

Quoted 5 lines
drivers/hwmon/lm90.c +5 -0
+	if (client->irq >= 0) {
+		INIT_WORK(&data->irq_work, lm90_irq_work);
+		dev_dbg(dev, "lm90 irq: %d\n", client->irq);
+		err = request_irq(client->irq, lm90_irq, IRQF_TRIGGER_LOW,
+				       "lm90", data);

request_threaded_irq or even better devm_request_threaded_irq would be better here.

yes, it's better to use request_threaded_irq here.
we can set lm90_irq() as interrupt handle, and set lm90_alert() as thread function.
I will try it on my next version.

Quoted 32 lines
Quoted 27 lines
drivers/hwmon/lm90.c +7 -0
+		if (err < 0) {
+			dev_err(dev, "cannot request interrupt\n");
+			goto exit_remove_files;
+		}
+	}
+
 	return 0;
 
 exit_remove_files:
@@ -1507,6 +1544,7 @@ static int lm90_remove(struct i2c_client *client)
 {
 	struct lm90_data *data = i2c_get_clientdata(client);
 
+	free_irq(client->irq, data);
 	hwmon_device_unregister(data->hwmon_dev);
 	lm90_remove_files(client, data);
 	lm90_restore_conf(client, data);
Signature · 7 lines
1.7.9.5


_______________________________________________
lm-sensors mailing list
lm-sensors at lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

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

#15 lore raw

RFC PATCH 3/9 hwmon: (lm90) add support to handle irq

In reply to Wei Ni
Quoted 6 lines · Wei Ni

Add support to handle irq. When the temperature touch the limit value, the driver can handle the interrupt.

drivers/hwmon/lm90.c +1 -0
 
+static void lm90_irq_work(struct work_struct *work)

...

drivers/hwmon/lm90.c static void lm90_init_client(struct i2c_client *client) · line 1441

Quoted 8 lines
drivers/hwmon/lm90.c +8 -0
+	enable_irq(client->irq);
+}
+
+static irqreturn_t lm90_irq(int irq, void *dev_id)
+{
+	struct lm90_data *data = dev_id;
+
+	disable_irq_nosync(irq);

I think IRQF_ONESHOT would enable you to remove the manual enable/disable_irq calls.

#16 lore raw

RFC PATCH 3/9 hwmon: (lm90) add support to handle irq

Quoted 8 lines · Stephen Warren
Quoted 6 lines · Wei Ni

Add support to handle irq. When the temperature touch the limit value, the driver can handle the interrupt.

drivers/hwmon/lm90.c +1 -0
 
+static void lm90_irq_work(struct work_struct *work)

...

Quoted 11 lines

drivers/hwmon/lm90.c static void lm90_init_client(struct i2c_client *client) · line 1441

Quoted 8 lines
drivers/hwmon/lm90.c +8 -0
+	enable_irq(client->irq);
+}
+
+static irqreturn_t lm90_irq(int irq, void *dev_id)
+{
+	struct lm90_data *data = dev_id;
+
+	disable_irq_nosync(irq);

I think IRQF_ONESHOT would enable you to remove the manual enable/disable_irq calls.

Oh, right. Using a threaded IRQ with IRQF_ONESHOT would make that happen transparently. Wasn't aware of that, thanks.

Alex.

author
#17 lore raw

RFC PATCH 3/9 hwmon: (lm90) add support to handle irq

Quoted 21 lines · Alex Courbot
Quoted 8 lines · Stephen Warren
Quoted 6 lines · Wei Ni

Add support to handle irq. When the temperature touch the limit value, the driver can handle the interrupt.

drivers/hwmon/lm90.c +1 -0
 
+static void lm90_irq_work(struct work_struct *work)

...

Quoted 11 lines

drivers/hwmon/lm90.c static void lm90_init_client(struct i2c_client *client) · line 1441

Quoted 8 lines
drivers/hwmon/lm90.c +8 -0
+	enable_irq(client->irq);
+}
+
+static irqreturn_t lm90_irq(int irq, void *dev_id)
+{
+	struct lm90_data *data = dev_id;
+
+	disable_irq_nosync(irq);

I think IRQF_ONESHOT would enable you to remove the manual enable/disable_irq calls.

Quoted 2 lines

Oh, right. Using a threaded IRQ with IRQF_ONESHOT would make that happen transparently. Wasn't aware of that, thanks.

Oh, yes, I will use it in my next version.

Thanks.
Wei.

Quoted 4 lines

Alex.

author
#18 lore raw

RFC PATCH 4/9 hwmon: (lm90) use macros for the indexes of temp8 and temp11

In reply to Wei Ni

Using macros for the indexes and nrs of temp8 and temp11. This make the code much clearer.

---
 drivers/hwmon/lm90.c |  179 ++++++++++++++++++++++++++++++++------------------
 1 file changed, 114 insertions(+), 65 deletions(-)
drivers/hwmon/lm90.c +114 -65
index 80311ef..de5a476 100644
@@ -298,6 +298,59 @@ static const struct lm90_params lm90_params[] = {
 };
 
 /*
+ * TEMP8 register index
+ */
+enum lm90_temp8_reg_index {
+	TEMP8_LOCAL_LOW = 0,	/* 0: local low limit */
+	TEMP8_LOCAL_HIGH,	/* 1: local high limit */
+	TEMP8_LOCAL_CRIT,	/* 2: local critical limit */
+	TEMP8_REMOTE_CRIT,	/* 3: remote critical limit */
+	TEMP8_LOCAL_EMERG,	/* 4: local emergency limit
+				 * (max6659 and max6695/96)
+				 */
+	TEMP8_REMOTE_EMERG,	/* 5: remote emergency limit
+				 * (max6659 and max6695/96)
+				 */
+	TEMP8_REMOTE2_CRIT,	/* 6: remote 2 critical limit
+				 * (max6695/96 only)
+				 */
+	TEMP8_REMOTE2_EMERG,	/* 7: remote 2 emergency limit
+				 * (max6695/96 only)
+				 */
+	TEMP8_REG_NUM
+};
+
+/*
+ * TEMP11 register index
+ */
+enum lm90_temp11_reg_index {
+	TEMP11_REMOTE_TEMP = 0,	/* 0: remote input */
+	TEMP11_REMOTE_LOW,	/* 1: remote low limit */
+	TEMP11_REMOTE_HIGH,	/* 2: remote high limit */
+	TEMP11_REMOTE_OFFSET,	/* 3: remote offset
+				 * (except max6646, max6657/58/59,
+				 *  and max6695/96)
+				 */
+	TEMP11_LOCAL_TEMP,	/* 4: local input */
+	TEMP11_REMOTE2_TEMP,	/* 5: remote 2 input (max6695/96 only) */
+	TEMP11_REMOTE2_LOW,	/* 6: remote 2 low limit (max6695/96 only) */
+	TEMP11_REMOTE2_HIGH,	/* 7: remote 2 high limit (max6695/96 only) */
+	TEMP11_REG_NUM
+};
+
+/*
+ * TEMP11 register NR
+ */
+enum lm90_temp11_reg_nr {
+	NR_CHAN_0_REMOTE_LOW = 0,	/* 0: channel 0, remote low limit */
+	NR_CHAN_0_REMOTE_HIGH,		/* 1: channel 0, remote high limit */
+	NR_CHAN_0_REMOTE_OFFSET,	/* 2: channel 0, remote offset */
+	NR_CHAN_1_REMOTE_LOW,		/* 3: channel 1, remote low limit */
+	NR_CHAN_1_REMOTE_HIGH,		/* 4: channel 1, remote high limit */
+	NR_NUM				/* number of the NRs for temp11 */
+};
+
+/*
  * Client data (each client gets its own)
  */
 
@@ -320,25 +373,8 @@ struct lm90_data {
 	u8 reg_local_ext;	/* local extension register offset */
 
 	/* registers values */
-	s8 temp8[8];	/* 0: local low limit
-			 * 1: local high limit
-			 * 2: local critical limit
-			 * 3: remote critical limit
-			 * 4: local emergency limit (max6659 and max6695/96)
-			 * 5: remote emergency limit (max6659 and max6695/96)
-			 * 6: remote 2 critical limit (max6695/96 only)
-			 * 7: remote 2 emergency limit (max6695/96 only)
-			 */
-	s16 temp11[8];	/* 0: remote input
-			 * 1: remote low limit
-			 * 2: remote high limit
-			 * 3: remote offset (except max6646, max6657/58/59,
-			 *		     and max6695/96)
-			 * 4: local input
-			 * 5: remote 2 input (max6695/96 only)
-			 * 6: remote 2 low limit (max6695/96 only)
-			 * 7: remote 2 high limit (max6695/96 only)
-			 */
+	s8 temp8[TEMP8_REG_NUM];
+	s16 temp11[TEMP11_REG_NUM];
 	u8 temp_hyst;
 	u16 alarms; /* bitvector (upper 8 bits for max6695/96) */
 };
@@ -480,37 +516,42 @@ static struct lm90_data *lm90_update_device(struct device *dev)
 		u8 alarms;
 
 		dev_dbg(&client->dev, "Updating lm90 data.\n");
-		lm90_read_reg(client, LM90_REG_R_LOCAL_LOW, &data->temp8[0]);
-		lm90_read_reg(client, LM90_REG_R_LOCAL_HIGH, &data->temp8[1]);
-		lm90_read_reg(client, LM90_REG_R_LOCAL_CRIT, &data->temp8[2]);
-		lm90_read_reg(client, LM90_REG_R_REMOTE_CRIT, &data->temp8[3]);
+		lm90_read_reg(client, LM90_REG_R_LOCAL_LOW,
+				&data->temp8[TEMP8_LOCAL_LOW]);
+		lm90_read_reg(client, LM90_REG_R_LOCAL_HIGH,
+				&data->temp8[TEMP8_LOCAL_HIGH]);
+		lm90_read_reg(client, LM90_REG_R_LOCAL_CRIT,
+				&data->temp8[TEMP8_LOCAL_CRIT]);
+		lm90_read_reg(client, LM90_REG_R_REMOTE_CRIT,
+				&data->temp8[TEMP8_REMOTE_CRIT]);
 		lm90_read_reg(client, LM90_REG_R_TCRIT_HYST, &data->temp_hyst);
 
 		if (data->reg_local_ext) {
 			lm90_read16(client, LM90_REG_R_LOCAL_TEMP,
 				    data->reg_local_ext,
-				    &data->temp11[4]);
+				    &data->temp11[TEMP11_LOCAL_TEMP]);
 		} else {
 			if (lm90_read_reg(client, LM90_REG_R_LOCAL_TEMP,
 					  &h) == 0)
-				data->temp11[4] = h << 8;
+				data->temp11[TEMP11_LOCAL_TEMP] = h << 8;
 		}
 		lm90_read16(client, LM90_REG_R_REMOTE_TEMPH,
-			    LM90_REG_R_REMOTE_TEMPL, &data->temp11[0]);
+			LM90_REG_R_REMOTE_TEMPL,
+			&data->temp11[TEMP11_REMOTE_TEMP]);
 
 		if (lm90_read_reg(client, LM90_REG_R_REMOTE_LOWH, &h) == 0) {
-			data->temp11[1] = h << 8;
+			data->temp11[TEMP11_REMOTE_LOW] = h << 8;
 			if ((data->flags & LM90_HAVE_REM_LIMIT_EXT)
 			 && lm90_read_reg(client, LM90_REG_R_REMOTE_LOWL,
 					  &l) == 0)
-				data->temp11[1] |= l;
+				data->temp11[TEMP11_REMOTE_LOW] |= l;
 		}
 		if (lm90_read_reg(client, LM90_REG_R_REMOTE_HIGHH, &h) == 0) {
-			data->temp11[2] = h << 8;
+			data->temp11[TEMP11_REMOTE_HIGH] = h << 8;
 			if ((data->flags & LM90_HAVE_REM_LIMIT_EXT)
 			 && lm90_read_reg(client, LM90_REG_R_REMOTE_HIGHL,
 					  &l) == 0)
-				data->temp11[2] |= l;
+				data->temp11[TEMP11_REMOTE_HIGH] |= l;
 		}
 
 		if (data->flags & LM90_HAVE_OFFSET) {
@@ -518,13 +559,14 @@ static struct lm90_data *lm90_update_device(struct device *dev)
 					  &h) == 0
 			 && lm90_read_reg(client, LM90_REG_R_REMOTE_OFFSL,
 					  &l) == 0)
-				data->temp11[3] = (h << 8) | l;
+				data->temp11[TEMP11_REMOTE_OFFSET] =
+								(h << 8) | l;
 		}
 		if (data->flags & LM90_HAVE_EMERGENCY) {
 			lm90_read_reg(client, MAX6659_REG_R_LOCAL_EMERG,
-				      &data->temp8[4]);
+				      &data->temp8[TEMP8_LOCAL_EMERG]);
 			lm90_read_reg(client, MAX6659_REG_R_REMOTE_EMERG,
-				      &data->temp8[5]);
+				      &data->temp8[TEMP8_REMOTE_EMERG]);
 		}
 		lm90_read_reg(client, LM90_REG_R_STATUS, &alarms);
 		data->alarms = alarms;	/* save as 16 bit value */
@@ -532,15 +574,16 @@ static struct lm90_data *lm90_update_device(struct device *dev)
 		if (data->kind == max6696) {
 			lm90_select_remote_channel(client, data, 1);
 			lm90_read_reg(client, LM90_REG_R_REMOTE_CRIT,
-				      &data->temp8[6]);
+				      &data->temp8[TEMP8_REMOTE2_CRIT]);
 			lm90_read_reg(client, MAX6659_REG_R_REMOTE_EMERG,
-				      &data->temp8[7]);
+				      &data->temp8[TEMP8_REMOTE2_EMERG]);
 			lm90_read16(client, LM90_REG_R_REMOTE_TEMPH,
-				    LM90_REG_R_REMOTE_TEMPL, &data->temp11[5]);
+					LM90_REG_R_REMOTE_TEMPL,
+					&data->temp11[TEMP11_REMOTE2_TEMP]);
 			if (!lm90_read_reg(client, LM90_REG_R_REMOTE_LOWH, &h))
-				data->temp11[6] = h << 8;
+				data->temp11[TEMP11_REMOTE2_LOW] = h << 8;
 			if (!lm90_read_reg(client, LM90_REG_R_REMOTE_HIGHH, &h))
-				data->temp11[7] = h << 8;
+				data->temp11[TEMP11_REMOTE2_HIGH] = h << 8;
 			lm90_select_remote_channel(client, data, 0);
 
 			if (!lm90_read_reg(client, MAX6696_REG_R_STATUS2,
@@ -735,7 +778,7 @@ static ssize_t show_temp8(struct device *dev, struct device_attribute *devattr,
 
 static void _set_temp8(struct device *dev, int index, long val)
 {
-	static const u8 reg[8] = {
+	static const u8 reg[TEMP8_REG_NUM] = {
 		LM90_REG_W_LOCAL_LOW,
 		LM90_REG_W_LOCAL_HIGH,
 		LM90_REG_W_LOCAL_CRIT,
@@ -818,7 +861,7 @@ static void _set_temp11(struct device *dev, int nr, int index, long val)
 		u8 high;
 		u8 low;
 		int channel;
-	} reg[5] = {
+	} reg[NR_NUM] = {
 		{ LM90_REG_W_REMOTE_LOWH, LM90_REG_W_REMOTE_LOWL, 0 },
 		{ LM90_REG_W_REMOTE_HIGHH, LM90_REG_W_REMOTE_HIGHL, 0 },
 		{ LM90_REG_W_REMOTE_OFFSH, LM90_REG_W_REMOTE_OFFSL, 0 },
@@ -909,11 +952,12 @@ static ssize_t set_temphyst(struct device *dev, struct device_attribute *dummy,
 
 	mutex_lock(&data->update_lock);
 	if (data->kind == adt7461)
-		temp = temp_from_u8_adt7461(data, data->temp8[2]);
+		temp = temp_from_u8_adt7461(data,
+					data->temp8[TEMP8_LOCAL_CRIT]);
 	else if (data->kind == max6646)
-		temp = temp_from_u8(data->temp8[2]);
+		temp = temp_from_u8(data->temp8[TEMP8_LOCAL_CRIT]);
 	else
-		temp = temp_from_s8(data->temp8[2]);
+		temp = temp_from_s8(data->temp8[TEMP8_LOCAL_CRIT]);
 
 	data->temp_hyst = hyst_to_reg(temp - val);
 	i2c_smbus_write_byte_data(client, LM90_REG_W_TCRIT_HYST,
@@ -967,25 +1011,28 @@ static ssize_t set_update_interval(struct device *dev,
 	return count;
 }
 
-static SENSOR_DEVICE_ATTR_2(temp1_input, S_IRUGO, show_temp11, NULL, 0, 4);
-static SENSOR_DEVICE_ATTR_2(temp2_input, S_IRUGO, show_temp11, NULL, 0, 0);
+static SENSOR_DEVICE_ATTR_2(temp1_input, S_IRUGO, show_temp11, NULL,
+	NR_CHAN_0_REMOTE_LOW, TEMP11_LOCAL_TEMP);
+static SENSOR_DEVICE_ATTR_2(temp2_input, S_IRUGO, show_temp11, NULL,
+	NR_CHAN_0_REMOTE_LOW, TEMP11_REMOTE_TEMP);
 static SENSOR_DEVICE_ATTR(temp1_min, S_IWUSR | S_IRUGO, show_temp8,
-	set_temp8, 0);
+	set_temp8, TEMP8_LOCAL_LOW);
 static SENSOR_DEVICE_ATTR_2(temp2_min, S_IWUSR | S_IRUGO, show_temp11,
-	set_temp11, 0, 1);
+	set_temp11, NR_CHAN_0_REMOTE_LOW, TEMP11_REMOTE_LOW);
 static SENSOR_DEVICE_ATTR(temp1_max, S_IWUSR | S_IRUGO, show_temp8,
-	set_temp8, 1);
+	set_temp8, TEMP8_LOCAL_HIGH);
 static SENSOR_DEVICE_ATTR_2(temp2_max, S_IWUSR | S_IRUGO, show_temp11,
-	set_temp11, 1, 2);
+	set_temp11, NR_CHAN_0_REMOTE_HIGH, TEMP11_REMOTE_HIGH);
 static SENSOR_DEVICE_ATTR(temp1_crit, S_IWUSR | S_IRUGO, show_temp8,
-	set_temp8, 2);
+	set_temp8, TEMP8_LOCAL_CRIT);
 static SENSOR_DEVICE_ATTR(temp2_crit, S_IWUSR | S_IRUGO, show_temp8,
-	set_temp8, 3);
+	set_temp8, TEMP8_REMOTE_CRIT);
 static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IWUSR | S_IRUGO, show_temphyst,
-	set_temphyst, 2);
-static SENSOR_DEVICE_ATTR(temp2_crit_hyst, S_IRUGO, show_temphyst, NULL, 3);
+	set_temphyst, TEMP8_LOCAL_CRIT);
+static SENSOR_DEVICE_ATTR(temp2_crit_hyst, S_IRUGO, show_temphyst, NULL,
+	TEMP8_REMOTE_CRIT);
 static SENSOR_DEVICE_ATTR_2(temp2_offset, S_IWUSR | S_IRUGO, show_temp11,
-	set_temp11, 2, 3);
+	set_temp11, NR_CHAN_0_REMOTE_OFFSET, TEMP11_REMOTE_OFFSET);
 
 /* Individual alarm files */
 static SENSOR_DEVICE_ATTR(temp1_crit_alarm, S_IRUGO, show_alarm, NULL, 0);
@@ -1033,13 +1080,13 @@ static const struct attribute_group lm90_group = {
  * Additional attributes for devices with emergency sensors
  */
 static SENSOR_DEVICE_ATTR(temp1_emergency, S_IWUSR | S_IRUGO, show_temp8,
-	set_temp8, 4);
+	set_temp8, TEMP8_LOCAL_EMERG);
 static SENSOR_DEVICE_ATTR(temp2_emergency, S_IWUSR | S_IRUGO, show_temp8,
-	set_temp8, 5);
+	set_temp8, TEMP8_REMOTE_EMERG);
 static SENSOR_DEVICE_ATTR(temp1_emergency_hyst, S_IRUGO, show_temphyst,
-			  NULL, 4);
+			  NULL, TEMP8_LOCAL_EMERG);
 static SENSOR_DEVICE_ATTR(temp2_emergency_hyst, S_IRUGO, show_temphyst,
-			  NULL, 5);
+			  NULL, TEMP8_REMOTE_EMERG);
 
 static struct attribute *lm90_emergency_attributes[] = {
 	&sensor_dev_attr_temp1_emergency.dev_attr.attr,
@@ -1069,18 +1116,20 @@ static const struct attribute_group lm90_emergency_alarm_group = {
 /*
  * Additional attributes for devices with 3 temperature sensors
  */
-static SENSOR_DEVICE_ATTR_2(temp3_input, S_IRUGO, show_temp11, NULL, 0, 5);
+static SENSOR_DEVICE_ATTR_2(temp3_input, S_IRUGO, show_temp11, NULL,
+	NR_CHAN_0_REMOTE_LOW, TEMP11_REMOTE2_TEMP);
 static SENSOR_DEVICE_ATTR_2(temp3_min, S_IWUSR | S_IRUGO, show_temp11,
-	set_temp11, 3, 6);
+	set_temp11, NR_CHAN_1_REMOTE_LOW, TEMP11_REMOTE2_LOW);
 static SENSOR_DEVICE_ATTR_2(temp3_max, S_IWUSR | S_IRUGO, show_temp11,
-	set_temp11, 4, 7);
+	set_temp11, NR_CHAN_1_REMOTE_HIGH, TEMP11_REMOTE2_HIGH);
 static SENSOR_DEVICE_ATTR(temp3_crit, S_IWUSR | S_IRUGO, show_temp8,
-	set_temp8, 6);
-static SENSOR_DEVICE_ATTR(temp3_crit_hyst, S_IRUGO, show_temphyst, NULL, 6);
+	set_temp8, TEMP8_REMOTE2_CRIT);
+static SENSOR_DEVICE_ATTR(temp3_crit_hyst, S_IRUGO, show_temphyst, NULL,
+	TEMP8_REMOTE2_CRIT);
 static SENSOR_DEVICE_ATTR(temp3_emergency, S_IWUSR | S_IRUGO, show_temp8,
-	set_temp8, 7);
+	set_temp8, TEMP8_REMOTE2_EMERG);
 static SENSOR_DEVICE_ATTR(temp3_emergency_hyst, S_IRUGO, show_temphyst,
-			  NULL, 7);
+			  NULL, TEMP8_REMOTE2_EMERG);
 
 static SENSOR_DEVICE_ATTR(temp3_crit_alarm, S_IRUGO, show_alarm, NULL, 9);
 static SENSOR_DEVICE_ATTR(temp3_fault, S_IRUGO, show_alarm, NULL, 10);
Signature · 1 line
1.7.9.5
#19 lore raw

RFC PATCH 4/9 hwmon: (lm90) use macros for the indexes of temp8 and temp11

Quoted 4 lines · Wei Ni

Using macros for the indexes and nrs of temp8 and temp11. This make the code much clearer.

Great patch. This makes the code much more readable. I also compiled lm90.o before and after this patch, and diff'ed the stripped binaries to make sure there was no error in the macro indexes. Both binaries were indentical, therefore:

Alex.

author
#20 lore raw

RFC PATCH 4/9 hwmon: (lm90) use macros for the indexes of temp8 and temp11

Quoted 6 lines · Alex Courbot
Quoted 4 lines · Wei Ni

Using macros for the indexes and nrs of temp8 and temp11. This make the code much clearer.

Quoted 8 lines

Great patch. This makes the code much more readable. I also compiled lm90.o before and after this patch, and diff'ed the stripped binaries to make sure there was no error in the macro indexes. Both binaries were indentical, therefore:

Great, thanks for your test, I will do it in my next version.

Wei.

Quoted 4 lines

Alex.

#21 lore raw

RFC PATCH 4/9 hwmon: (lm90) use macros for the indexes of temp8 and temp11

In reply to Wei Ni
Quoted 2 lines · Wei Ni

Using macros for the indexes and nrs of temp8 and temp11. This make the code much clearer.

Sorry to nit-pick, but those aren't macros, but rather they're enums.

The patch looks good though.

author
#22 lore raw

RFC PATCH 4/9 hwmon: (lm90) use macros for the indexes of temp8 and temp11

Quoted 5 lines · Stephen Warren
Quoted 2 lines · Wei Ni

Using macros for the indexes and nrs of temp8 and temp11. This make the code much clearer.

Sorry to nit-pick, but those aren't macros, but rather they're enums.

Yes, you are right, it's better to write "enums".
I will change it.

Wei.

Quoted 3 lines

The patch looks good though.

author
#23 lore raw

RFC PATCH 5/9 Thermal: Support using dt node to get sensor

In reply to Wei Ni

Add functions to support using dt node with args to get sensor.

---
 drivers/thermal/thermal_sys.c |   30 ++++++++++++++++++++++++++++++
 include/linux/thermal.h       |    9 +++++++++
 2 files changed, 39 insertions(+)
drivers/thermal/thermal_sys.c +30 -0
index e284b67..b5bedab 100644
@@ -2183,6 +2183,28 @@ struct thermal_sensor *get_sensor_by_name(const char *name)
 }
 EXPORT_SYMBOL(get_sensor_by_name);
 
+struct thermal_sensor *get_sensor_by_node(struct node_args *np_args)
+{
+	struct thermal_sensor *pos;
+	struct thermal_sensor *ts = NULL;
+	struct node_args *args;
+
+	mutex_lock(&sensor_list_lock);
+	for_each_thermal_sensor(pos) {
+		args = &pos->np_args;
+		if (args->np) {
+			if ((args->np == np_args->np) &&
+			   (args->index == np_args->index)) {
+				ts = pos;
+				break;
+			}
+		}
+	}
+	mutex_unlock(&sensor_list_lock);
+	return ts;
+}
+EXPORT_SYMBOL(get_sensor_by_node);
+
 static int create_map_attrs(struct thermal_zone *tz, int indx)
 {
 	int ret, i;
@@ -2368,6 +2390,7 @@ EXPORT_SYMBOL(add_sensor_trip_info);
  * @devdata:	private device data
  */
 struct thermal_sensor *thermal_sensor_register(const char *name, int count,
+			struct node_args *np_args,
 			struct thermal_sensor_ops *ops, void *devdata)
 {
 	struct thermal_sensor *ts;
@@ -2412,6 +2435,13 @@ struct thermal_sensor *thermal_sensor_register(const char *name, int count,
 			goto exit_temp;
 	}
 
+	if (np_args) {
+		ts->np_args.np = np_args->np;
+		ts->np_args.index = np_args->index;
+	} else {
+		ts->np_args.np = NULL;
+	}
+
 	/* Add this sensor to the global list of sensors */
 	mutex_lock(&sensor_list_lock);
 	list_add_tail(&ts->node, &thermal_sensor_list);
include/linux/thermal.h +9 -0
index 4389599..b560ffa 100644
@@ -113,6 +113,12 @@ enum {
 };
 #define THERMAL_GENL_CMD_MAX (__THERMAL_GENL_CMD_MAX - 1)
 
+/* support to parse node with args to get sensor or cooling device */
+struct node_args {
+	struct device_node *np;
+	int index;
+};
+
 struct thermal_zone_device_ops {
 	int (*bind) (struct thermal_zone_device *,
 		     struct thermal_cooling_device *);
@@ -223,6 +229,7 @@ struct thermal_sensor {
 	void *devdata;
 	struct idr idr;
 	struct device device;
+	struct node_args np_args;
 	struct list_head node;
 	struct thermal_sensor_ops *ops;
 	struct thermal_attr *thresh_attrs;
@@ -350,6 +357,7 @@ int thermal_register_governor(struct thermal_governor *);
 void thermal_unregister_governor(struct thermal_governor *);
 
 struct thermal_sensor *thermal_sensor_register(const char *, int,
+				struct node_args *,
 				struct thermal_sensor_ops *, void *);
 void thermal_sensor_unregister(struct thermal_sensor *);
 
@@ -357,6 +365,7 @@ struct thermal_zone *create_thermal_zone(const char *, void *);
 void remove_thermal_zone(struct thermal_zone *);
 int add_sensor_to_zone(struct thermal_zone *, struct thermal_sensor *);
 struct thermal_sensor *get_sensor_by_name(const char *);
+struct thermal_sensor *get_sensor_by_node(struct node_args *);
 
 int add_cdev_to_zone(struct thermal_zone *, struct thermal_cooling_device *);
 struct thermal_cooling_device *get_cdev_by_name(const char *);
Signature · 1 line
1.7.9.5
#24 lore raw

RFC PATCH 5/9 Thermal: Support using dt node to get sensor

Quoted 1 line · Wei Ni

Add functions to support using dt node with args to get sensor.

You need to write a device tree binding document to explain this.

drivers/thermal/thermal_sys.c struct thermal_sensor *get_sensor_by_name(const char *name) · line 2188

Quoted 5 lines
drivers/thermal/thermal_sys.c +3 -0
 
+struct thermal_sensor *get_sensor_by_node(struct node_args *np_args)
+{
+	struct thermal_sensor *pos;

"pos" isn't a great variable name. Why not use "sensor", or just the "ts" variable you have right below?

drivers/thermal/thermal_sys.c struct thermal_sensor *get_sensor_by_name(const char *name) · line 2199

Quoted 11 lines
drivers/thermal/thermal_sys.c +11 -0
+	struct thermal_sensor *ts = NULL;
+	struct node_args *args;
+
+	mutex_lock(&sensor_list_lock);
+	for_each_thermal_sensor(pos) {
+		args = &pos->np_args;
+		if (args->np) {
+			if ((args->np == np_args->np) &&
+			   (args->index == np_args->index)) {
+				ts = pos;
+				break;

Replace those 2 lines with "goto out;".

drivers/thermal/thermal_sys.c struct thermal_sensor *get_sensor_by_name(const char *name) · line 2202

Quoted 3 lines
drivers/thermal/thermal_sys.c +3 -0
+			}
+		}
+	}

here, add:

	ts = NULL;

out:

That way, you can use "ts" as the loop iteration variable.

This whole patch rather assumes that all DT nodes can identify their exposed thermal sensors using an index in a single DT cell. That's not very flexible. All other DT bindings work like this:

Provider of a service indicates how many DT cells are in the object (GPIO, IRQ, thermal sensors) specifier:

sensor1: lm90 at 1c {
	...
	#thermal-sensor-cells = <1>;
};

Each consumer of a service imports it by referencing it:

thermal-zone {
	...
	sensors = <&sensor1 0>;
};

The driver for LM90 provides an "of_xlate" function which receives a struct of_phandle_args and outputs/returns whatever Linux-internal identification/representation of the object is required. For example, see:

Quoted 1 line

include/linux/pwm.h:161: struct pwm_device * (*of_xlate)(struct pwm_chip *pc,

This allows each providing object's DT binding to define its own value of #thermal-sensor-cells, as suited for its own requirements, and allows each driver to implement the mapping from DT to internal ID in whatever way is necessary.

Now, many bindings/drivers might just end up using a common simple implementation. That's why functions such as of_pwm_simple_xlate() or of_gpio_simple_xlate() exist.

author
#25 lore raw

RFC PATCH 5/9 Thermal: Support using dt node to get sensor

Quoted 5 lines · Stephen Warren
Quoted 1 line · Wei Ni

Add functions to support using dt node with args to get sensor.

You need to write a device tree binding document to explain this.

Quoted 8 lines

drivers/thermal/thermal_sys.c struct thermal_sensor *get_sensor_by_name(const char *name) · line 2188

Quoted 5 lines
drivers/thermal/thermal_sys.c +3 -0
 
+struct thermal_sensor *get_sensor_by_node(struct node_args *np_args)
+{
+	struct thermal_sensor *pos;

"pos" isn't a great variable name. Why not use "sensor", or just the "ts" variable you have right below?

Oh, yes, it can just use "ts" simply. I didn't consider it, thanks.

Quoted 51 lines

drivers/thermal/thermal_sys.c struct thermal_sensor *get_sensor_by_name(const char *name) · line 2199

Quoted 12 lines
drivers/thermal/thermal_sys.c +11 -0
+	struct thermal_sensor *ts = NULL;
+	struct node_args *args;
+
+	mutex_lock(&sensor_list_lock);
+	for_each_thermal_sensor(pos) {
+		args = &pos->np_args;
+		if (args->np) {
+			if ((args->np == np_args->np) &&
+			   (args->index == np_args->index)) {
+				ts = pos;
+				break;

Replace those 2 lines with "goto out;".

drivers/thermal/thermal_sys.c struct thermal_sensor *get_sensor_by_name(const char *name) · line 2202

Quoted 3 lines
drivers/thermal/thermal_sys.c +3 -0
+			}
+		}
+	}

here, add:

	ts = NULL;

out:

That way, you can use "ts" as the loop iteration variable.

This whole patch rather assumes that all DT nodes can identify their exposed thermal sensors using an index in a single DT cell. That's not very flexible. All other DT bindings work like this:

Provider of a service indicates how many DT cells are in the object (GPIO, IRQ, thermal sensors) specifier:

sensor1: lm90 at 1c {
	...
	#thermal-sensor-cells = <1>;
};

Each consumer of a service imports it by referencing it:

thermal-zone {
	...
	sensors = <&sensor1 0>;
};

The driver for LM90 provides an "of_xlate" function which receives a struct of_phandle_args and outputs/returns whatever Linux-internal identification/representation of the object is required. For example, see:

Quoted 1 line

include/linux/pwm.h:161: struct pwm_device * (*of_xlate)(struct pwm_chip *pc,

Quoted 8 lines

This allows each providing object's DT binding to define its own value of #thermal-sensor-cells, as suited for its own requirements, and allows each driver to implement the mapping from DT to internal ID in whatever way is necessary.

Now, many bindings/drivers might just end up using a common simple implementation. That's why functions such as of_pwm_simple_xlate() or of_gpio_simple_xlate() exist.

It looks like the "of_xlate" can handle the sensor identification. I'm not familiar with this function, I will look into it and try to use. Thanks for your comments.

Wei.

Quoted 1 line
author
#26 lore raw

RFC PATCH 6/9 hwmon: (lm90) Register to the thermal framework

In reply to Wei Ni

Register the remote sensor to the thermal framework. It can support to show the temperature and read/write threshold.

---
 arch/arm/boot/dts/tegra30-cardhu.dtsi |    1 +
 drivers/hwmon/lm90.c                  |  182 ++++++++++++++++++++++++++++++++-
 2 files changed, 182 insertions(+), 1 deletion(-)
arch/arm/boot/dts/tegra30-cardhu.dtsi +1 -0
index 15ad1ad..3f6ab89 100644
@@ -279,6 +279,7 @@
 			reg = <0x4c>;
 			interrupt-parent = <&gpio>;
 			interrupts = <226 0x08>; /* gpio PCC2 */
+			#sensor-cells = <1>;
 		};
 	};
 
drivers/hwmon/lm90.c +181 -1

229 lines not on this page · raw message

Signature · 1 line
1.7.9.5
maintainer
#27 lore raw

RFC PATCH 6/9 hwmon: (lm90) Register to the thermal framework

Quoted 85 lines · Wei Ni

Register the remote sensor to the thermal framework. It can support to show the temperature and read/write threshold.

---
 arch/arm/boot/dts/tegra30-cardhu.dtsi |    1 +
 drivers/hwmon/lm90.c                  |  182 ++++++++++++++++++++++++++++++++-
 2 files changed, 182 insertions(+), 1 deletion(-)
arch/arm/boot/dts/tegra30-cardhu.dtsi +1 -0
index 15ad1ad..3f6ab89 100644
@@ -279,6 +279,7 @@
 			reg = <0x4c>;
 			interrupt-parent = <&gpio>;
 			interrupts = <226 0x08>; /* gpio PCC2 */
+			#sensor-cells = <1>;
 		};
 	};
 
drivers/hwmon/lm90.c +35 -0
index de5a476..0abdedc 100644
@@ -91,6 +91,7 @@
 #include <linux/sysfs.h>
 #include <linux/interrupt.h>
 #include <linux/of_irq.h>
+#include <linux/thermal.h>
 
 /*
  * Addresses to scan
@@ -182,6 +183,15 @@ enum chips { lm90, adm1032, lm99, lm86, max6657, max6659, adt7461, max6680,
 #define LM90_HAVE_BROKEN_ALERT	(1 << 7) /* Broken alert		*/
 
 /*
+ * Thermal framework
+ */
+enum lm90_thresholds {
+	LM90_LOW_THRESHOLDS = 0,	/* threshold 0: low limits */
+	LM90_HIGH_THRESHOLDS,		/* threshold 1: high limits */
+	LM90_NUM_THRESHOLDS
+};
+
+/*
  * Driver data (common to all clients)
  */
 
@@ -377,6 +387,9 @@ struct lm90_data {
 	s16 temp11[TEMP11_REG_NUM];
 	u8 temp_hyst;
 	u16 alarms; /* bitvector (upper 8 bits for max6695/96) */
+
+	struct thermal_sensor *ts_remote;
+	struct thermal_sensor *ts_local;
 };
 
 /*
@@ -1493,12 +1506,151 @@ static irqreturn_t lm90_irq(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
+static int lm90_read_remote_temp(struct thermal_sensor *ts, long *temp)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+
+	_show_temp11(dev, TEMP11_REMOTE_TEMP, (int *)temp);
+
+	return 0;
+}
+
+static int lm90_read_remote_threshold(struct thermal_sensor *ts, int th_index,
+					long *val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* remote low limit */
+		index = TEMP11_REMOTE_LOW;
+		break;

drivers/hwmon/lm90.c static irqreturn_t lm90_irq(int irq, void *dev_id) · line 1540

Quoted 10 lines
drivers/hwmon/lm90.c +10 -0
+	case LM90_HIGH_THRESHOLDS:
+		/* remote high limit */
+		index = TEMP11_REMOTE_HIGH;
+		break;
+	default:
+		dev_err(dev, "read remote threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_show_temp11(dev, index, (int *)val);

Typecasting a pointer like this doesn't work. Try this on a big-endian system with sizeof(int) != sizeof(long). Note that you would not have the problem if you would pass the value instead of the pointer.

Quoted 35 lines
drivers/hwmon/lm90.c +35 -0
+
+	return 0;
+}
+
+static int lm90_write_remote_threshold(struct thermal_sensor *ts, int th_index,
+					long val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int nr, index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* remote low limit */
+		nr = NR_CHAN_0_REMOTE_LOW;
+		index = TEMP11_REMOTE_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* remote high limit */
+		nr = NR_CHAN_0_REMOTE_HIGH;
+		index = TEMP11_REMOTE_HIGH;
+		break;
+	default:
+		dev_err(dev, "write remote threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_set_temp11(dev, nr, index, val);
+
+	return 0;
+}
+
+static struct thermal_sensor_ops remote_ops = {
+	.get_temp = lm90_read_remote_temp,
+	.get_threshold = lm90_read_remote_threshold,

drivers/hwmon/lm90.c static irqreturn_t lm90_irq(int irq, void *dev_id) · line 1585

Quoted 10 lines
drivers/hwmon/lm90.c +10 -0
+	.set_threshold = lm90_write_remote_threshold,
+};
+
+static int lm90_read_local_temp(struct thermal_sensor *ts, long *temp)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+
+	_show_temp11(dev, TEMP11_LOCAL_TEMP, (int *)temp);
+

Same here and everywhere else.

Quoted 94 lines
drivers/hwmon/lm90.c +77 -1
+	return 0;
+}
+
+static int lm90_read_local_threshold(struct thermal_sensor *ts, int th_index,
+					long *val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* local low limit */
+		index = TEMP8_LOCAL_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* local high limit */
+		index = TEMP8_LOCAL_HIGH;
+		break;
+	default:
+		dev_err(dev, "read local threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_show_temp8(dev, index, (int *)val);
+
+	return 0;
+}
+
+static int lm90_write_local_threshold(struct thermal_sensor *ts, int th_index,
+					long val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* local low limit */
+		index = TEMP8_LOCAL_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* local high limit */
+		index = TEMP8_LOCAL_HIGH;
+		break;
+	default:
+		dev_err(dev, "write local threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_set_temp8(dev, index, val);
+
+	return 0;
+}
+
+static struct thermal_sensor_ops local_ops = {
+	.get_temp = lm90_read_local_temp,
+	.get_threshold = lm90_read_local_threshold,
+	.set_threshold = lm90_write_local_threshold,
+};
+
 static int lm90_probe(struct i2c_client *client,
 		      const struct i2c_device_id *id)
 {
 	struct device *dev = &client->dev;
 	struct i2c_adapter *adapter = to_i2c_adapter(dev->parent);
 	struct lm90_data *data;
+	struct node_args np_args;
 	int err;
 
 	data = devm_kzalloc(&client->dev, sizeof(struct lm90_data), GFP_KERNEL);
@@ -1576,12 +1728,38 @@ static int lm90_probe(struct i2c_client *client,
 				       "lm90", data);
 		if (err < 0) {
 			dev_err(dev, "cannot request interrupt\n");
-			goto exit_remove_files;
+			goto exit_unregister_hwmon;
 		}
 	}
 
+	np_args.np = dev->of_node;
+	np_args.index = 0;
+	data->ts_remote = thermal_sensor_register("lm90_remote",
+						LM90_NUM_THRESHOLDS,
+						&np_args,
+						&remote_ops, client);
+	if (IS_ERR(data->ts_remote)) {
+		dev_err(dev, "cannot register sensor to thermal framework\n");
+		err = -EINVAL;
+		goto exit_unregister_hwmon;
+	}
+
+	np_args.index = 1;
+	data->ts_local = thermal_sensor_register("lm90_local",

drivers/hwmon/lm90.c static int lm90_probe(struct i2c_client *client, · line 1758

Quoted 10 lines
drivers/hwmon/lm90.c +10 -0
+						LM90_NUM_THRESHOLDS,
+						&np_args,
+						&local_ops, client);
+
+	if (IS_ERR(data->ts_local)) {
+		dev_err(dev, "cannot register sensor to thermal framework\n");
+		err = -EINVAL;
+		goto exit_unregister_hwmon;
+	}
+

How about the second remote sensor ? Granted, not all chips supported by this driver have it, but if we are adding this we might as well add it for all sensors on all chips.

Quoted 25 lines
drivers/hwmon/lm90.c +4 -0
 	return 0;
 
+exit_unregister_hwmon:
+	hwmon_device_unregister(data->hwmon_dev);
 exit_remove_files:
 	lm90_remove_files(client, data);
 exit_restore:
@@ -1594,6 +1772,8 @@ static int lm90_remove(struct i2c_client *client)
 	struct lm90_data *data = i2c_get_clientdata(client);
 
 	free_irq(client->irq, data);
+	thermal_sensor_unregister(data->ts_remote);
+	thermal_sensor_unregister(data->ts_local);
 	hwmon_device_unregister(data->hwmon_dev);
 	lm90_remove_files(client, data);
 	lm90_restore_conf(client, data);
Signature · 7 lines
1.7.9.5


_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
author
#28 lore raw

RFC PATCH 6/9 hwmon: (lm90) Register to the thermal framework

Quoted 92 lines · Guenter Roeck
Quoted 91 lines · Wei Ni

Register the remote sensor to the thermal framework. It can support to show the temperature and read/write threshold.

---
 arch/arm/boot/dts/tegra30-cardhu.dtsi |    1 +
 drivers/hwmon/lm90.c                  |  182 ++++++++++++++++++++++++++++++++-
 2 files changed, 182 insertions(+), 1 deletion(-)
arch/arm/boot/dts/tegra30-cardhu.dtsi +1 -0
index 15ad1ad..3f6ab89 100644
@@ -279,6 +279,7 @@
 			reg = <0x4c>;
 			interrupt-parent = <&gpio>;
 			interrupts = <226 0x08>; /* gpio PCC2 */
+			#sensor-cells = <1>;
 		};
 	};
 
drivers/hwmon/lm90.c +41 -0
index de5a476..0abdedc 100644
@@ -91,6 +91,7 @@
 #include <linux/sysfs.h>
 #include <linux/interrupt.h>
 #include <linux/of_irq.h>
+#include <linux/thermal.h>
 
 /*
  * Addresses to scan
@@ -182,6 +183,15 @@ enum chips { lm90, adm1032, lm99, lm86, max6657, max6659, adt7461, max6680,
 #define LM90_HAVE_BROKEN_ALERT	(1 << 7) /* Broken alert		*/
 
 /*
+ * Thermal framework
+ */
+enum lm90_thresholds {
+	LM90_LOW_THRESHOLDS = 0,	/* threshold 0: low limits */
+	LM90_HIGH_THRESHOLDS,		/* threshold 1: high limits */
+	LM90_NUM_THRESHOLDS
+};
+
+/*
  * Driver data (common to all clients)
  */
 
@@ -377,6 +387,9 @@ struct lm90_data {
 	s16 temp11[TEMP11_REG_NUM];
 	u8 temp_hyst;
 	u16 alarms; /* bitvector (upper 8 bits for max6695/96) */
+
+	struct thermal_sensor *ts_remote;
+	struct thermal_sensor *ts_local;
 };
 
 /*
@@ -1493,12 +1506,151 @@ static irqreturn_t lm90_irq(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
+static int lm90_read_remote_temp(struct thermal_sensor *ts, long *temp)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+
+	_show_temp11(dev, TEMP11_REMOTE_TEMP, (int *)temp);
+
+	return 0;
+}
+
+static int lm90_read_remote_threshold(struct thermal_sensor *ts, int th_index,
+					long *val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* remote low limit */
+		index = TEMP11_REMOTE_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* remote high limit */
+		index = TEMP11_REMOTE_HIGH;
+		break;
+	default:
+		dev_err(dev, "read remote threshold failed.\n");
Quoted 8 lines

drivers/hwmon/lm90.c static irqreturn_t lm90_irq(int irq, void *dev_id) · line 1540

Quoted 4 lines
drivers/hwmon/lm90.c +4 -0
+		return -EINVAL;
+	}
+
+	_show_temp11(dev, index, (int *)val);

Typecasting a pointer like this doesn't work. Try this on a big-endian system with sizeof(int) != sizeof(long). Note that you would not have the problem if you would pass the value instead of the pointer.

Yes, you are right, it will cause problems.
I will return the value for _show_temp11() and _show_temp8().

Quoted 148 lines
Quoted 36 lines
drivers/hwmon/lm90.c +35 -0
+
+	return 0;
+}
+
+static int lm90_write_remote_threshold(struct thermal_sensor *ts, int th_index,
+					long val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int nr, index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* remote low limit */
+		nr = NR_CHAN_0_REMOTE_LOW;
+		index = TEMP11_REMOTE_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* remote high limit */
+		nr = NR_CHAN_0_REMOTE_HIGH;
+		index = TEMP11_REMOTE_HIGH;
+		break;
+	default:
+		dev_err(dev, "write remote threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_set_temp11(dev, nr, index, val);
+
+	return 0;
+}
+
+static struct thermal_sensor_ops remote_ops = {
+	.get_temp = lm90_read_remote_temp,
+	.get_threshold = lm90_read_remote_threshold,

drivers/hwmon/lm90.c static irqreturn_t lm90_irq(int irq, void *dev_id) · line 1585

Quoted 10 lines
drivers/hwmon/lm90.c +10 -0
+	.set_threshold = lm90_write_remote_threshold,
+};
+
+static int lm90_read_local_temp(struct thermal_sensor *ts, long *temp)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+
+	_show_temp11(dev, TEMP11_LOCAL_TEMP, (int *)temp);
+

Same here and everywhere else.

Quoted 100 lines
drivers/hwmon/lm90.c +83 -1
+	return 0;
+}
+
+static int lm90_read_local_threshold(struct thermal_sensor *ts, int th_index,
+					long *val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* local low limit */
+		index = TEMP8_LOCAL_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* local high limit */
+		index = TEMP8_LOCAL_HIGH;
+		break;
+	default:
+		dev_err(dev, "read local threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_show_temp8(dev, index, (int *)val);
+
+	return 0;
+}
+
+static int lm90_write_local_threshold(struct thermal_sensor *ts, int th_index,
+					long val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* local low limit */
+		index = TEMP8_LOCAL_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* local high limit */
+		index = TEMP8_LOCAL_HIGH;
+		break;
+	default:
+		dev_err(dev, "write local threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_set_temp8(dev, index, val);
+
+	return 0;
+}
+
+static struct thermal_sensor_ops local_ops = {
+	.get_temp = lm90_read_local_temp,
+	.get_threshold = lm90_read_local_threshold,
+	.set_threshold = lm90_write_local_threshold,
+};
+
 static int lm90_probe(struct i2c_client *client,
 		      const struct i2c_device_id *id)
 {
 	struct device *dev = &client->dev;
 	struct i2c_adapter *adapter = to_i2c_adapter(dev->parent);
 	struct lm90_data *data;
+	struct node_args np_args;
 	int err;
 
 	data = devm_kzalloc(&client->dev, sizeof(struct lm90_data), GFP_KERNEL);
@@ -1576,12 +1728,38 @@ static int lm90_probe(struct i2c_client *client,
 				       "lm90", data);
 		if (err < 0) {
 			dev_err(dev, "cannot request interrupt\n");
-			goto exit_remove_files;
+			goto exit_unregister_hwmon;
 		}
 	}
 
+	np_args.np = dev->of_node;
+	np_args.index = 0;
+	data->ts_remote = thermal_sensor_register("lm90_remote",
+						LM90_NUM_THRESHOLDS,
+						&np_args,
+						&remote_ops, client);
+	if (IS_ERR(data->ts_remote)) {
+		dev_err(dev, "cannot register sensor to thermal framework\n");
+		err = -EINVAL;
+		goto exit_unregister_hwmon;
+	}
+
+	np_args.index = 1;
+	data->ts_local = thermal_sensor_register("lm90_local",
+						LM90_NUM_THRESHOLDS,
+						&np_args,
+						&local_ops, client);
+
+	if (IS_ERR(data->ts_local)) {
+		dev_err(dev, "cannot register sensor to thermal framework\n");
Quoted 8 lines

drivers/hwmon/lm90.c static int lm90_probe(struct i2c_client *client, · line 1746

Quoted 4 lines
drivers/hwmon/lm90.c +4 -0
+		err = -EINVAL;
+		goto exit_unregister_hwmon;
+	}
+

How about the second remote sensor ? Granted, not all chips supported by this driver have it, but if we are adding this we might as well add it for all sensors on all chips.

Ok, I will improve it.

Quoted 26 lines
Quoted 26 lines
drivers/hwmon/lm90.c +4 -0
 	return 0;
 
+exit_unregister_hwmon:
+	hwmon_device_unregister(data->hwmon_dev);
 exit_remove_files:
 	lm90_remove_files(client, data);
 exit_restore:
@@ -1594,6 +1772,8 @@ static int lm90_remove(struct i2c_client *client)
 	struct lm90_data *data = i2c_get_clientdata(client);
 
 	free_irq(client->irq, data);
+	thermal_sensor_unregister(data->ts_remote);
+	thermal_sensor_unregister(data->ts_local);
 	hwmon_device_unregister(data->hwmon_dev);
 	lm90_remove_files(client, data);
 	lm90_restore_conf(client, data);
Signature · 7 lines
1.7.9.5


_______________________________________________
lm-sensors mailing list
lm-sensors at lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
#29 lore raw

RFC PATCH 6/9 hwmon: (lm90) Register to the thermal framework

In reply to Wei Ni
Quoted 8 lines · Wei Ni

Register the remote sensor to the thermal framework. It can support to show the temperature and read/write threshold.

---
  arch/arm/boot/dts/tegra30-cardhu.dtsi |    1 +
  drivers/hwmon/lm90.c                  |  182 ++++++++++++++++++++++++++++++++-
  2 files changed, 182 insertions(+), 1 deletion(-)

Making changes to a driver *and* a board file in the same patch? I think this should be separated, and the board file change preferably squashed with the first patch of this series, and moved right after this one.

Quoted 52 lines
arch/arm/boot/dts/tegra30-cardhu.dtsi +1 -0
index 15ad1ad..3f6ab89 100644
@@ -279,6 +279,7 @@
  			reg = <0x4c>;
  			interrupt-parent = <&gpio>;
  			interrupts = <226 0x08>; /* gpio PCC2 */
+			#sensor-cells = <1>;
  		};
  	};
 
drivers/hwmon/lm90.c +13 -0
index de5a476..0abdedc 100644
@@ -91,6 +91,7 @@
  #include <linux/sysfs.h>
  #include <linux/interrupt.h>
  #include <linux/of_irq.h>
+#include <linux/thermal.h>
 
  /*
   * Addresses to scan
@@ -182,6 +183,15 @@ enum chips { lm90, adm1032, lm99, lm86, max6657, max6659, adt7461, max6680,
  #define LM90_HAVE_BROKEN_ALERT	(1 << 7) /* Broken alert		*/
 
  /*
+ * Thermal framework
+ */
+enum lm90_thresholds {
+	LM90_LOW_THRESHOLDS = 0,	/* threshold 0: low limits */
+	LM90_HIGH_THRESHOLDS,		/* threshold 1: high limits */
+	LM90_NUM_THRESHOLDS
+};
+
+/*
   * Driver data (common to all clients)
   */
 
@@ -377,6 +387,9 @@ struct lm90_data {
  	s16 temp11[TEMP11_REG_NUM];
  	u8 temp_hyst;
  	u16 alarms; /* bitvector (upper 8 bits for max6695/96) */
+
+	struct thermal_sensor *ts_remote;
+	struct thermal_sensor *ts_local;
  };
 
  /*

drivers/hwmon/lm90.c static irqreturn_t lm90_irq(int irq, void *dev_id) · line 1514

Quoted 9 lines
drivers/hwmon/lm90.c +6 -0
@@ -1493,12 +1506,151 @@ static irqreturn_t lm90_irq(int irq, void *dev_id)
  	return IRQ_HANDLED;
  }
 
+static int lm90_read_remote_temp(struct thermal_sensor *ts, long *temp)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+
+	_show_temp11(dev, TEMP11_REMOTE_TEMP, (int *)temp);

As Guenter pointed, this might break. Since you introduced _show_temp11 in a previous patch, you should revise it to take a long * as third argument (or even better, return a long). Or if you cannot do that for some reason, use a temporary int and affect temp properly (*temp = temp_int).

Quoted 80 lines
drivers/hwmon/lm90.c +80 -0
+
+	return 0;
+}
+
+static int lm90_read_remote_threshold(struct thermal_sensor *ts, int th_index,
+					long *val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* remote low limit */
+		index = TEMP11_REMOTE_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* remote high limit */
+		index = TEMP11_REMOTE_HIGH;
+		break;
+	default:
+		dev_err(dev, "read remote threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_show_temp11(dev, index, (int *)val);
+
+	return 0;
+}
+
+static int lm90_write_remote_threshold(struct thermal_sensor *ts, int th_index,
+					long val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int nr, index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* remote low limit */
+		nr = NR_CHAN_0_REMOTE_LOW;
+		index = TEMP11_REMOTE_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* remote high limit */
+		nr = NR_CHAN_0_REMOTE_HIGH;
+		index = TEMP11_REMOTE_HIGH;
+		break;
+	default:
+		dev_err(dev, "write remote threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_set_temp11(dev, nr, index, val);
+
+	return 0;
+}
+
+static struct thermal_sensor_ops remote_ops = {
+	.get_temp = lm90_read_remote_temp,
+	.get_threshold = lm90_read_remote_threshold,
+	.set_threshold = lm90_write_remote_threshold,
+};
+
+static int lm90_read_local_temp(struct thermal_sensor *ts, long *temp)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+
+	_show_temp11(dev, TEMP11_LOCAL_TEMP, (int *)temp);
+
+	return 0;
+}
+
+static int lm90_read_local_threshold(struct thermal_sensor *ts, int th_index,
+					long *val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;

drivers/hwmon/lm90.c static irqreturn_t lm90_irq(int irq, void *dev_id) · line 1604

Quoted 10 lines
drivers/hwmon/lm90.c +10 -0
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* local low limit */
+		index = TEMP8_LOCAL_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* local high limit */
+		index = TEMP8_LOCAL_HIGH;
+		break;

I think the comments are unneeded here, the macro name should be explicit enough.

Quoted 60 lines
drivers/hwmon/lm90.c +44 -1
+	default:
+		dev_err(dev, "read local threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_show_temp8(dev, index, (int *)val);
+
+	return 0;
+}
+
+static int lm90_write_local_threshold(struct thermal_sensor *ts, int th_index,
+					long val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* local low limit */
+		index = TEMP8_LOCAL_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* local high limit */
+		index = TEMP8_LOCAL_HIGH;
+		break;
+	default:
+		dev_err(dev, "write local threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_set_temp8(dev, index, val);
+
+	return 0;
+}
+
+static struct thermal_sensor_ops local_ops = {
+	.get_temp = lm90_read_local_temp,
+	.get_threshold = lm90_read_local_threshold,
+	.set_threshold = lm90_write_local_threshold,
+};
+
  static int lm90_probe(struct i2c_client *client,
  		      const struct i2c_device_id *id)
  {
  	struct device *dev = &client->dev;
  	struct i2c_adapter *adapter = to_i2c_adapter(dev->parent);
  	struct lm90_data *data;
+	struct node_args np_args;
  	int err;
 
  	data = devm_kzalloc(&client->dev, sizeof(struct lm90_data), GFP_KERNEL);
@@ -1576,12 +1728,38 @@ static int lm90_probe(struct i2c_client *client,
  				       "lm90", data);
  		if (err < 0) {
  			dev_err(dev, "cannot request interrupt\n");
-			goto exit_remove_files;
+			goto exit_unregister_hwmon;
  		}
  	}

drivers/hwmon/lm90.c static int lm90_probe(struct i2c_client *client, · line 1743

Quoted 10 lines
drivers/hwmon/lm90.c +9 -0
 
+	np_args.np = dev->of_node;
+	np_args.index = 0;
+	data->ts_remote = thermal_sensor_register("lm90_remote",
+						LM90_NUM_THRESHOLDS,
+						&np_args,
+						&remote_ops, client);
+	if (IS_ERR(data->ts_remote)) {
+		dev_err(dev, "cannot register sensor to thermal framework\n");
+		err = -EINVAL;

When don't you return the error code provided by thermal_sensor_register, e.g. err = PTR_ERR(data->ts_remote) ?

drivers/hwmon/lm90.c static int lm90_probe(struct i2c_client *client, · line 1755

Quoted 12 lines
drivers/hwmon/lm90.c +12 -0
+		goto exit_unregister_hwmon;
+	}
+
+	np_args.index = 1;
+	data->ts_local = thermal_sensor_register("lm90_local",
+						LM90_NUM_THRESHOLDS,
+						&np_args,
+						&local_ops, client);
+
+	if (IS_ERR(data->ts_local)) {
+		dev_err(dev, "cannot register sensor to thermal framework\n");
+		err = -EINVAL;

Same thing here.

Quoted 11 lines
drivers/hwmon/lm90.c +5 -0
+		goto exit_unregister_hwmon;
+	}
+
  	return 0;
 
+exit_unregister_hwmon:
+	hwmon_device_unregister(data->hwmon_dev);
  exit_remove_files:
  	lm90_remove_files(client, data);
  exit_restore:

drivers/hwmon/lm90.c static int lm90_remove(struct i2c_client *client) · line 1776

Quoted 5 lines
drivers/hwmon/lm90.c +2 -0
@@ -1594,6 +1772,8 @@ static int lm90_remove(struct i2c_client *client)
  	struct lm90_data *data = i2c_get_clientdata(client);
 
  	free_irq(client->irq, data);
+	thermal_sensor_unregister(data->ts_remote);
+	thermal_sensor_unregister(data->ts_local);

Ideally you would unregister your sensors in the reverse order they have been registered, but I'm being picky here.

Alex.

author
#30 lore raw

RFC PATCH 6/9 hwmon: (lm90) Register to the thermal framework

Quoted 69 lines · Alex Courbot
Quoted 8 lines · Wei Ni

Register the remote sensor to the thermal framework. It can support to show the temperature and read/write threshold.

---
  arch/arm/boot/dts/tegra30-cardhu.dtsi |    1 +
  drivers/hwmon/lm90.c                  |  182 ++++++++++++++++++++++++++++++++-
  2 files changed, 182 insertions(+), 1 deletion(-)

Making changes to a driver *and* a board file in the same patch? I think this should be separated, and the board file change preferably squashed with the first patch of this series, and moved right after this one.

Quoted 55 lines
arch/arm/boot/dts/tegra30-cardhu.dtsi +1 -0
index 15ad1ad..3f6ab89 100644
@@ -279,6 +279,7 @@
  			reg = <0x4c>;
  			interrupt-parent = <&gpio>;
  			interrupts = <226 0x08>; /* gpio PCC2 */
+			#sensor-cells = <1>;
  		};
  	};
 
drivers/hwmon/lm90.c +13 -0
index de5a476..0abdedc 100644
@@ -91,6 +91,7 @@
  #include <linux/sysfs.h>
  #include <linux/interrupt.h>
  #include <linux/of_irq.h>
+#include <linux/thermal.h>
 
  /*
   * Addresses to scan
@@ -182,6 +183,15 @@ enum chips { lm90, adm1032, lm99, lm86, max6657, max6659, adt7461, max6680,
  #define LM90_HAVE_BROKEN_ALERT	(1 << 7) /* Broken alert		*/
 
  /*
+ * Thermal framework
+ */
+enum lm90_thresholds {
+	LM90_LOW_THRESHOLDS = 0,	/* threshold 0: low limits */
+	LM90_HIGH_THRESHOLDS,		/* threshold 1: high limits */
+	LM90_NUM_THRESHOLDS
+};
+
+/*
   * Driver data (common to all clients)
   */
 
@@ -377,6 +387,9 @@ struct lm90_data {
  	s16 temp11[TEMP11_REG_NUM];
  	u8 temp_hyst;
  	u16 alarms; /* bitvector (upper 8 bits for max6695/96) */
+
+	struct thermal_sensor *ts_remote;
+	struct thermal_sensor *ts_local;
  };
 
  /*
@@ -1493,12 +1506,151 @@ static irqreturn_t lm90_irq(int irq, void *dev_id)
  	return IRQ_HANDLED;
  }
 
Quoted 12 lines

drivers/hwmon/lm90.c static irqreturn_t lm90_irq(int irq, void *dev_id) · line 1514

Quoted 6 lines
drivers/hwmon/lm90.c +6 -0
+static int lm90_read_remote_temp(struct thermal_sensor *ts, long *temp)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+
+	_show_temp11(dev, TEMP11_REMOTE_TEMP, (int *)temp);

As Guenter pointed, this might break. Since you introduced _show_temp11 in a previous patch, you should revise it to take a long * as third argument (or even better, return a long). Or if you cannot do that for some reason, use a temporary int and affect temp properly (*temp = temp_int).

yes, the pointer will cause problems here.
I will follow Guenter suggestion to return the value simply for _show_temp11 and _show_temp8

Quoted 86 lines
Quoted 86 lines
drivers/hwmon/lm90.c +85 -0
+
+	return 0;
+}
+
+static int lm90_read_remote_threshold(struct thermal_sensor *ts, int th_index,
+					long *val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* remote low limit */
+		index = TEMP11_REMOTE_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* remote high limit */
+		index = TEMP11_REMOTE_HIGH;
+		break;
+	default:
+		dev_err(dev, "read remote threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_show_temp11(dev, index, (int *)val);
+
+	return 0;
+}
+
+static int lm90_write_remote_threshold(struct thermal_sensor *ts, int th_index,
+					long val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int nr, index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* remote low limit */
+		nr = NR_CHAN_0_REMOTE_LOW;
+		index = TEMP11_REMOTE_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* remote high limit */
+		nr = NR_CHAN_0_REMOTE_HIGH;
+		index = TEMP11_REMOTE_HIGH;
+		break;
+	default:
+		dev_err(dev, "write remote threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_set_temp11(dev, nr, index, val);
+
+	return 0;
+}
+
+static struct thermal_sensor_ops remote_ops = {
+	.get_temp = lm90_read_remote_temp,
+	.get_threshold = lm90_read_remote_threshold,
+	.set_threshold = lm90_write_remote_threshold,
+};
+
+static int lm90_read_local_temp(struct thermal_sensor *ts, long *temp)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+
+	_show_temp11(dev, TEMP11_LOCAL_TEMP, (int *)temp);
+
+	return 0;
+}
+
+static int lm90_read_local_threshold(struct thermal_sensor *ts, int th_index,
+					long *val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* local low limit */
+		index = TEMP8_LOCAL_LOW;
Quoted 8 lines

drivers/hwmon/lm90.c static irqreturn_t lm90_irq(int irq, void *dev_id) · line 1604

Quoted 5 lines
drivers/hwmon/lm90.c +5 -0
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* local high limit */
+		index = TEMP8_LOCAL_HIGH;
+		break;

I think the comments are unneeded here, the macro name should be explicit enough.

OK.

Quoted 62 lines
Quoted 62 lines
drivers/hwmon/lm90.c +44 -1
+	default:
+		dev_err(dev, "read local threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_show_temp8(dev, index, (int *)val);
+
+	return 0;
+}
+
+static int lm90_write_local_threshold(struct thermal_sensor *ts, int th_index,
+					long val)
+{
+	struct i2c_client *client = ts->devdata;
+	struct device *dev = &client->dev;
+	int index;
+
+	switch (th_index) {
+	case LM90_LOW_THRESHOLDS:
+		/* local low limit */
+		index = TEMP8_LOCAL_LOW;
+		break;
+	case LM90_HIGH_THRESHOLDS:
+		/* local high limit */
+		index = TEMP8_LOCAL_HIGH;
+		break;
+	default:
+		dev_err(dev, "write local threshold failed.\n");
+		return -EINVAL;
+	}
+
+	_set_temp8(dev, index, val);
+
+	return 0;
+}
+
+static struct thermal_sensor_ops local_ops = {
+	.get_temp = lm90_read_local_temp,
+	.get_threshold = lm90_read_local_threshold,
+	.set_threshold = lm90_write_local_threshold,
+};
+
  static int lm90_probe(struct i2c_client *client,
  		      const struct i2c_device_id *id)
  {
  	struct device *dev = &client->dev;
  	struct i2c_adapter *adapter = to_i2c_adapter(dev->parent);
  	struct lm90_data *data;
+	struct node_args np_args;
  	int err;
 
  	data = devm_kzalloc(&client->dev, sizeof(struct lm90_data), GFP_KERNEL);
@@ -1576,12 +1728,38 @@ static int lm90_probe(struct i2c_client *client,
  				       "lm90", data);
  		if (err < 0) {
  			dev_err(dev, "cannot request interrupt\n");
-			goto exit_remove_files;
+			goto exit_unregister_hwmon;
  		}
  	}
 
Quoted 12 lines

drivers/hwmon/lm90.c static int lm90_probe(struct i2c_client *client, · line 1743

Quoted 9 lines
drivers/hwmon/lm90.c +9 -0
+	np_args.np = dev->of_node;
+	np_args.index = 0;
+	data->ts_remote = thermal_sensor_register("lm90_remote",
+						LM90_NUM_THRESHOLDS,
+						&np_args,
+						&remote_ops, client);
+	if (IS_ERR(data->ts_remote)) {
+		dev_err(dev, "cannot register sensor to thermal framework\n");
+		err = -EINVAL;

When don't you return the error code provided by thermal_sensor_register, e.g. err = PTR_ERR(data->ts_remote) ?

I didn't consider it, I will change it.

Quoted 29 lines

drivers/hwmon/lm90.c static int lm90_probe(struct i2c_client *client, · line 1755

Quoted 13 lines
drivers/hwmon/lm90.c +12 -0
+		goto exit_unregister_hwmon;
+	}
+
+	np_args.index = 1;
+	data->ts_local = thermal_sensor_register("lm90_local",
+						LM90_NUM_THRESHOLDS,
+						&np_args,
+						&local_ops, client);
+
+	if (IS_ERR(data->ts_local)) {
+		dev_err(dev, "cannot register sensor to thermal framework\n");
+		err = -EINVAL;

Same thing here.

Quoted 13 lines
drivers/hwmon/lm90.c +5 -0
+		goto exit_unregister_hwmon;
+	}
+
  	return 0;
 
+exit_unregister_hwmon:
+	hwmon_device_unregister(data->hwmon_dev);
  exit_remove_files:
  	lm90_remove_files(client, data);
  exit_restore:
@@ -1594,6 +1772,8 @@ static int lm90_remove(struct i2c_client *client)
  	struct lm90_data *data = i2c_get_clientdata(client);
 
Quoted 6 lines

drivers/hwmon/lm90.c static int lm90_remove(struct i2c_client *client) · line 1776

Quoted 3 lines
drivers/hwmon/lm90.c +2 -0
  	free_irq(client->irq, data);
+	thermal_sensor_unregister(data->ts_remote);
+	thermal_sensor_unregister(data->ts_local);

Ideally you would unregister your sensors in the reverse order they have been registered, but I'm being picky here.

Yes, it's better in reverse order.
I really appreciate you reviewing my patches so carefully :)

Wei.

Quoted 3 lines

Alex.

#31 lore raw

RFC PATCH 6/9 hwmon: (lm90) Register to the thermal framework

In reply to Wei Ni
Quoted 5 lines · Wei Ni

Register the remote sensor to the thermal framework. It can support to show the temperature and read/write threshold.

 arch/arm/boot/dts/tegra30-cardhu.dtsi |    1 +
 drivers/hwmon/lm90.c                  |  182 ++++++++++++++++++++++++++++++++-

There shouldn't be any need to edit both those files in the same patch. This patch is making multiple separate logical changes, which should be separate patches.

arch/arm/boot/dts/tegra30-cardhu.dtsi

Quoted 3 lines
arch/arm/boot/dts/tegra30-cardhu.dtsi +1 -0
 
+			#sensor-cells = <1>;

Again, in order to allow that, you need some documentation in Documentation/devicetree/bindings/. Related to that, you need to Cc the devicetree maintainers on at least the patches that add that documentation.

Quoted 3 lines
drivers/hwmon/lm90.c +0 -0
@@ -1576,12 +1728,38 @@ static int lm90_probe(struct i2c_client *client,

...

drivers/hwmon/lm90.c static int lm90_probe(struct i2c_client *client, · line 1736

Quoted 2 lines
drivers/hwmon/lm90.c +2 -0
+	np_args.np = dev->of_node;
+	np_args.index = 0;

As I mentioned before, that'd be better replaced by an explicit of_xlate function.

author
#32 lore raw

RFC PATCH 6/9 hwmon: (lm90) Register to the thermal framework

Quoted 10 lines · Stephen Warren
Quoted 5 lines · Wei Ni

Register the remote sensor to the thermal framework. It can support to show the temperature and read/write threshold.

 arch/arm/boot/dts/tegra30-cardhu.dtsi |    1 +
 drivers/hwmon/lm90.c                  |  182 ++++++++++++++++++++++++++++++++-

There shouldn't be any need to edit both those files in the same patch. This patch is making multiple separate logical changes, which should be separate patches.

I will follow Matthew's suggestion to split this serial patch into 3 separate sets.

Quoted 8 lines

arch/arm/boot/dts/tegra30-cardhu.dtsi

Quoted 4 lines
arch/arm/boot/dts/tegra30-cardhu.dtsi +1 -0
 
+			#sensor-cells = <1>;

Again, in order to allow that, you need some documentation in Documentation/devicetree/bindings/. Related to that, you need to Cc the devicetree maintainers on at least the patches that add that documentation.

Ok, I will do it.

Quoted 10 lines
Quoted 4 lines
drivers/hwmon/lm90.c +0 -0
@@ -1576,12 +1728,38 @@ static int lm90_probe(struct i2c_client *client,

...

drivers/hwmon/lm90.c static int lm90_probe(struct i2c_client *client, · line 1736

Quoted 2 lines
drivers/hwmon/lm90.c +2 -0
+	np_args.np = dev->of_node;
+	np_args.index = 0;

As I mentioned before, that'd be better replaced by an explicit of_xlate function.

Yes, I will try to use this func.

Quoted 5 lines

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

author
#33 lore raw

RFC PATCH 7/9 thermal: tegra30: add tegra30 thermal driver

In reply to Wei Ni

dd Tegra30 thermal driver support. It create thermal zone with thermal sensors and cooling device to participate in the linux thermal management.

---
 drivers/thermal/Kconfig          |    9 ++
 drivers/thermal/Makefile         |    1 +
 drivers/thermal/tegra3_thermal.c |  289 ++++++++++++++++++++++++++++++++++++++
 3 files changed, 299 insertions(+)
 create mode 100644 drivers/thermal/tegra3_thermal.c
drivers/thermal/Kconfig +9 -0
index eadef5b..2403681 100644
@@ -141,6 +141,15 @@ config INTEL_POWERCLAMP
 	  enforce idle time which results in more package C-state residency. The
 	  user interface is exposed via generic thermal framework.
 
+config TEGRA30_THERMAL
+	tristate "Tegra30 thermal driver"
+	depends on ARCH_TEGRA
+	help
+	  Select this to enable the Tegra30 thermal driver. Adds Tegra30 thermal
+	  implementation according to the thermal management framework. Create
+	  thermal zone with thermal sensors and cooling device to participate
+	  in the linux thermal management.
+
 config THERMAL_TEST
 	tristate "test driver"
 	help
drivers/thermal/Makefile +1 -0
index ee0f687..de0b411 100644
@@ -19,6 +19,7 @@ obj-$(CONFIG_EXYNOS_THERMAL)	+= exynos_thermal.o
 obj-$(CONFIG_DB8500_THERMAL)	+= db8500_thermal.o
 obj-$(CONFIG_DB8500_CPUFREQ_COOLING)	+= db8500_cpufreq_cooling.o
 obj-$(CONFIG_INTEL_POWERCLAMP)	+= intel_powerclamp.o
+obj-$(CONFIG_TEGRA30_THERMAL)	+= tegra3_thermal.o
 
 # dummy driver for testing
 obj-$(CONFIG_THERMAL_TEST)	+= thermal_test.o
drivers/thermal/tegra3_thermal.c new file +289 -0

289 lines not on this page · raw message

Signature · 1 line
1.7.9.5
#34 lore raw

RFC PATCH 7/9 thermal: tegra30: add tegra30 thermal driver

drivers/thermal/tegra3_thermal.c

Quoted 6 lines · Wei Ni

dd Tegra30 thermal driver support. It create thermal zone with thermal sensors and cooling device to participate in the linux thermal management.

drivers/thermal/tegra3_thermal.c +1 -0
 
+struct tegra_thermal_data {

That's not really "thermal data", but that "thermal zone" or "thermal device" itself.

drivers/thermal/tegra3_thermal.c line 89

Quoted 10 lines
drivers/thermal/tegra3_thermal.c +10 -0
+static int
+tegra_throttle_set_cur_state(struct thermal_cooling_device *cdev,
+				unsigned long cur_state)
+{
+	struct balanced_throttle *bthrot = cdev->devdata;
+	int index;
+
+	mutex_lock(&cpu_throttle_lock);
+
+	/* TODO: we will handle the dvfs here */

That seems like rather a large TODO. We don't have any DVFS support for Tegra yet. Is it even worth moving forward with this driver without that?

drivers/thermal/tegra3_thermal.c line 106

Quoted 2 lines
drivers/thermal/tegra3_thermal.c +2 -0
+struct thermal_cooling_device *balanced_throttle_register(
+		struct balanced_throttle *bthrot, char *type)

What does "balanced" mean here; isn't balanced a governor, whereas this driver is simply providing the implementation that an arbitrary governor would use to do its will?

This function also appears to be rather generically named. A consistent function/symbol-name prefix of e.g. "tegra30_thermal" would be useful throughout the file.

drivers/thermal/tegra3_thermal.c line 120

Quoted 2 lines
drivers/thermal/tegra3_thermal.c +2 -0
+static struct tegra_thermal_data * __devinit thermal_tegra_dt_parse_pdata(
+						struct platform_device *pdev)

I'd like some slight re-structing here.

Tegra only supports device tree now; no board files (upstream at least, which is all that's relevant here). Hence, none of the Tegra drivers support platform data at all after the last round of cleanup patches I sent on Friday (most of which haven't been checked in yet). Hence, this function is not parsing platform data out of DT, it's plain parsing device tree.

So, can you move the devm_kzalloc of the tdata into probe() right at the start, rename this function tegra30_thermal_parse_dt(), and call it right after allocating "tdata" (which also is more like "tdev" or "tzone" perhaps).

Downstream if you still need to support platform data, you can simply copy the platform data into "tdev" rather than calling tegra30_thermal_parse_dt(); a simple patch to carry.

drivers/thermal/tegra3_thermal.c line 136

Quoted 6 lines
drivers/thermal/tegra3_thermal.c +6 -0
+	tdata = devm_kzalloc(&pdev->dev, sizeof(*tdata), GFP_KERNEL);
+	if (!tdata) {
+		dev_err(&pdev->dev, "Can't allocate platform data\n");
+		return NULL;
+	}
+	memset(tdata, 0, sizeof(*tdata));

That last line is kinda the whole point of k*z*alloc...

drivers/thermal/tegra3_thermal.c line 145

Quoted 9 lines
drivers/thermal/tegra3_thermal.c +9 -0
+
+	ret = of_parse_phandle_with_args(np, "sensors", "#sensor-cells", 0,
+					&args);
+	if (ret) {
+		dev_err(&pdev->dev, "Can't get sensor.\n");
+		return NULL;
+	}
+	tdata->np_args.np = args.np;
+	tdata->np_args.index = args.args[0];

That lookup should be implemented as part of the thermal core. It shouldn't be duplicated in every single driver.

drivers/thermal/tegra3_thermal.c line 149

Quoted 3 lines
drivers/thermal/tegra3_thermal.c +3 -0
+	ret = of_property_read_u32(np, "passive-delay", &val);
+	if (!ret)
+		tdata->passive_delay = val;

The DT binding documentation doesn't say this property is optional. If you intend it to be, the documentation should say so, and specify what the default is if the property is missing.

drivers/thermal/tegra3_thermal.c line 157

Quoted 7 lines
drivers/thermal/tegra3_thermal.c +7 -0
+	ret = of_property_read_u32(np, "num-passive-trips", &val);
+	if (!ret)
+		tdata->trip_ext.num_passive_trips = val;
+
+	if (tdata->trip_ext.num_passive_trips) {
+		tdata->trip_ext.passive_trips = devm_kzalloc(&pdev->dev,
+						sizeof(int) * val, GFP_KERNEL);

DT cells are specifically U32 not int. You probably want the driver to use U32 instead of int to make 100% sure the range matches.

drivers/thermal/tegra3_thermal.c line 161

Quoted 4 lines
drivers/thermal/tegra3_thermal.c +4 -0
+
+		of_property_read_u32_array(np, "passive-trips",
+				(u32 *)(tdata->trip_ext.passive_trips),
+				tdata->trip_ext.num_passive_trips);

... and then you wouldn't need this cast!

drivers/thermal/tegra3_thermal.c line 183

Quoted 5 lines
drivers/thermal/tegra3_thermal.c +4 -0
+	}
 
+	of_property_read_u32_array(np, "throt-tab",
+				(u32 *)(&tdata->tj_throttle.throt_tab),
+				tdata->tj_throttle.throt_tab_size * 2);

What about error-handling that API call, and all the others?

drivers/thermal/tegra3_thermal.c line 190

Quoted 3 lines
drivers/thermal/tegra3_thermal.c +3 -0
+static int tegra30_thermal_probe(struct platform_device *pdev)
+{
+	struct tegra_thermal_data *pdata = pdev->dev.platform_data;

Lets not support platform data at all upstream.

drivers/thermal/tegra3_thermal.c line 212

Quoted 2 lines
drivers/thermal/tegra3_thermal.c +2 -0
+	/* Register cooling device */
+	cdev = balanced_throttle_register(&pdata->tj_throttle, "cdev_throttle");

Is this picking specific policy ("balanced")? Should it be? What is "cdev_throttle"?

author
#35 lore raw

RFC PATCH 7/9 thermal: tegra30: add tegra30 thermal driver

Quoted 11 lines · Stephen Warren

drivers/thermal/tegra3_thermal.c

Quoted 6 lines · Wei Ni

dd Tegra30 thermal driver support. It create thermal zone with thermal sensors and cooling device to participate in the linux thermal management.

drivers/thermal/tegra3_thermal.c +1 -0
 
+struct tegra_thermal_data {

That's not really "thermal data", but that "thermal zone" or "thermal device" itself.

Quoted 13 lines

drivers/thermal/tegra3_thermal.c line 89

Quoted 10 lines
drivers/thermal/tegra3_thermal.c +10 -0
+static int
+tegra_throttle_set_cur_state(struct thermal_cooling_device *cdev,
+				unsigned long cur_state)
+{
+	struct balanced_throttle *bthrot = cdev->devdata;
+	int index;
+
+	mutex_lock(&cpu_throttle_lock);
+
+	/* TODO: we will handle the dvfs here */

That seems like rather a large TODO. We don't have any DVFS support for Tegra yet. Is it even worth moving forward with this driver without that?

I think it's better to split this file into two driver, one is for the thermal sensor, it can read/write temperature. And the other one is for the cooling device, something like tegra3_cooling .c, it will call dvfs interface to handle throttling when dvfs is ready.
And by now, we can focus on this tegra3_thermal.c.

Quoted 11 lines

drivers/thermal/tegra3_thermal.c line 106

Quoted 3 lines
drivers/thermal/tegra3_thermal.c +2 -0
+struct thermal_cooling_device *balanced_throttle_register(
+		struct balanced_throttle *bthrot, char *type)

What does "balanced" mean here; isn't balanced a governor, whereas this driver is simply providing the implementation that an arbitrary governor would use to do its will?

This function also appears to be rather generically named. A consistent function/symbol-name prefix of e.g. "tegra30_thermal" would be useful throughout the file.

I port these codes from our downstream kernel, I will move them to the new tegra3_cooling.c, and since the dvfs is not ready, I will remain simplest codes, and remove the "balanced".

Quoted 13 lines

drivers/thermal/tegra3_thermal.c line 120

Quoted 3 lines
drivers/thermal/tegra3_thermal.c +2 -0
+static struct tegra_thermal_data * __devinit thermal_tegra_dt_parse_pdata(
+						struct platform_device *pdev)

I'd like some slight re-structing here.

Tegra only supports device tree now; no board files (upstream at least, which is all that's relevant here). Hence, none of the Tegra drivers support platform data at all after the last round of cleanup patches I sent on Friday (most of which haven't been checked in yet). Hence, this function is not parsing platform data out of DT, it's plain parsing device tree.

Quoted 4 lines

So, can you move the devm_kzalloc of the tdata into probe() right at the start, rename this function tegra30_thermal_parse_dt(), and call it right after allocating "tdata" (which also is more like "tdev" or "tzone" perhaps).

Got it, I will change it.

Quoted 5 lines

Downstream if you still need to support platform data, you can simply copy the platform data into "tdev" rather than calling tegra30_thermal_parse_dt(); a simple patch to carry.

Quoted 8 lines

drivers/thermal/tegra3_thermal.c line 136

Quoted 6 lines
drivers/thermal/tegra3_thermal.c +6 -0
+	tdata = devm_kzalloc(&pdev->dev, sizeof(*tdata), GFP_KERNEL);
+	if (!tdata) {
+		dev_err(&pdev->dev, "Can't allocate platform data\n");
+		return NULL;
+	}
+	memset(tdata, 0, sizeof(*tdata));

That last line is kinda the whole point of k*z*alloc...

Yes, I forgot to remove it.

Quoted 14 lines

drivers/thermal/tegra3_thermal.c line 145

Quoted 10 lines
drivers/thermal/tegra3_thermal.c +9 -0
+
+	ret = of_parse_phandle_with_args(np, "sensors", "#sensor-cells", 0,
+					&args);
+	if (ret) {
+		dev_err(&pdev->dev, "Can't get sensor.\n");
+		return NULL;
+	}
+	tdata->np_args.np = args.np;
+	tdata->np_args.index = args.args[0];

That lookup should be implemented as part of the thermal core. It shouldn't be duplicated in every single driver.

Quoted 7 lines

drivers/thermal/tegra3_thermal.c line 149

Quoted 3 lines
drivers/thermal/tegra3_thermal.c +3 -0
+	ret = of_property_read_u32(np, "passive-delay", &val);
+	if (!ret)
+		tdata->passive_delay = val;

The DT binding documentation doesn't say this property is optional. If you intend it to be, the documentation should say so, and specify what the default is if the property is missing.

Ok, I will change it.

Quoted 11 lines

drivers/thermal/tegra3_thermal.c line 157

Quoted 8 lines
drivers/thermal/tegra3_thermal.c +7 -0
+	ret = of_property_read_u32(np, "num-passive-trips", &val);
+	if (!ret)
+		tdata->trip_ext.num_passive_trips = val;
+
+	if (tdata->trip_ext.num_passive_trips) {
+		tdata->trip_ext.passive_trips = devm_kzalloc(&pdev->dev,
+						sizeof(int) * val, GFP_KERNEL);

DT cells are specifically U32 not int. You probably want the driver to use U32 instead of int to make 100% sure the range matches.

Ok.

Quoted 8 lines

drivers/thermal/tegra3_thermal.c line 161

Quoted 5 lines
drivers/thermal/tegra3_thermal.c +4 -0
+
+		of_property_read_u32_array(np, "passive-trips",
+				(u32 *)(tdata->trip_ext.passive_trips),
+				tdata->trip_ext.num_passive_trips);

... and then you wouldn't need this cast!

Quoted 7 lines

drivers/thermal/tegra3_thermal.c line 183

Quoted 5 lines
drivers/thermal/tegra3_thermal.c +4 -0
+	}
 
+	of_property_read_u32_array(np, "throt-tab",
+				(u32 *)(&tdata->tj_throttle.throt_tab),
+				tdata->tj_throttle.throt_tab_size * 2);

What about error-handling that API call, and all the others?

I will add it.

Quoted 6 lines

drivers/thermal/tegra3_thermal.c line 190

Quoted 4 lines
drivers/thermal/tegra3_thermal.c +3 -0
+static int tegra30_thermal_probe(struct platform_device *pdev)
+{
+	struct tegra_thermal_data *pdata = pdev->dev.platform_data;

Lets not support platform data at all upstream.

Got it, thanks.

Quoted 6 lines

drivers/thermal/tegra3_thermal.c line 212

Quoted 3 lines
drivers/thermal/tegra3_thermal.c +2 -0
+	/* Register cooling device */
+	cdev = balanced_throttle_register(&pdata->tj_throttle, "cdev_throttle");

Is this picking specific policy ("balanced")? Should it be? What is "cdev_throttle"?

These codes are ported from our downstream kernel, it may handle multiple cooling device, so it named as "balanced", but in here we didn't use it, so it's better to remove them, and remain simplest codes. I will change it as I said above.

Quoted 5 lines

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

#36 lore raw

RFC PATCH 7/9 thermal: tegra30: add tegra30 thermal driver

In reply to Wei Ni
Quoted 1 line · Wei Ni

+static struct tegra_thermal_data * __devinit thermal_tegra_dt_parse_pdata(

__dev* no longer exists.

drivers/thermal/tegra3_thermal.c line 136

Quoted 6 lines
drivers/thermal/tegra3_thermal.c new file +6 -0
+	tdata = devm_kzalloc(&pdev->dev, sizeof(*tdata), GFP_KERNEL);
+	if (!tdata) {
+		dev_err(&pdev->dev, "Can't allocate platform data\n");
+		return NULL;
+	}
+	memset(tdata, 0, sizeof(*tdata));

Useless memset. k*z*alloc already zeros the memory before returning.

drivers/thermal/tegra3_thermal.c line 190

Quoted 3 lines
drivers/thermal/tegra3_thermal.c new file +3 -0
+static int tegra30_thermal_probe(struct platform_device *pdev)
+{
+	struct tegra_thermal_data *pdata = pdev->dev.platform_data;

You read pdata here....

drivers/thermal/tegra3_thermal.c line 196

Quoted 6 lines
drivers/thermal/tegra3_thermal.c new file +6 -0
+	struct thermal_zone *tz;
+	struct thermal_sensor *ts;
+	static struct thermal_cooling_device *cdev;
+	int ret;
+
+	pdata = thermal_tegra_dt_parse_pdata(pdev);

and immediately overwrite it here.

drivers/thermal/tegra3_thermal.c line 209

Quoted 13 lines
drivers/thermal/tegra3_thermal.c new file +13 -0
+	if (!pdata) {
+		dev_err(&pdev->dev, "Get platform data failed.\n");
+		return -EINVAL;
+	}
+
+	/* Create a thermal zone */
+	tz = create_thermal_zone("tz_tegra", NULL);
+	if (!tz) {
+		dev_err(&pdev->dev, "Create thermal_zone failed.\n");
+		return -EINVAL;
+	}
+
+	pdata->tz = tz;

This isn't how you deal with driver data. Set driver data against a platform device using platform_set_drvdata(pdev, tz).

drivers/thermal/tegra3_thermal.c line 255

Quoted 3 lines
drivers/thermal/tegra3_thermal.c new file +3 -0
+static int tegra30_thermal_remove(struct platform_device *pdev)
+{
+	struct tegra_thermal_data *pdata = pdev->dev.platform_data;

and use platform_get_drvdata() here - and don't use pdata->tz.

	struct struct thermal_zone *tz = platform_get_drvdata(pdev);
author
#37 lore raw

RFC PATCH 7/9 thermal: tegra30: add tegra30 thermal driver

Quoted 4 lines · Russell King - ARM Linux
Quoted 1 line · Wei Ni

+static struct tegra_thermal_data * __devinit thermal_tegra_dt_parse_pdata(

__dev* no longer exists.

Ok, I will change it.

Quoted 9 lines

drivers/thermal/tegra3_thermal.c line 136

Quoted 7 lines
drivers/thermal/tegra3_thermal.c new file +6 -0
+	tdata = devm_kzalloc(&pdev->dev, sizeof(*tdata), GFP_KERNEL);
+	if (!tdata) {
+		dev_err(&pdev->dev, "Can't allocate platform data\n");
+		return NULL;
+	}
+	memset(tdata, 0, sizeof(*tdata));

Useless memset. k*z*alloc already zeros the memory before returning.

Yes, I forgot to remove this line.

Quoted 16 lines

drivers/thermal/tegra3_thermal.c line 190

Quoted 4 lines
drivers/thermal/tegra3_thermal.c new file +3 -0
+static int tegra30_thermal_probe(struct platform_device *pdev)
+{
+	struct tegra_thermal_data *pdata = pdev->dev.platform_data;

You read pdata here....

drivers/thermal/tegra3_thermal.c line 196

Quoted 6 lines
drivers/thermal/tegra3_thermal.c new file +6 -0
+	struct thermal_zone *tz;
+	struct thermal_sensor *ts;
+	static struct thermal_cooling_device *cdev;
+	int ret;
+
+	pdata = thermal_tegra_dt_parse_pdata(pdev);

and immediately overwrite it here.

Quoted 16 lines

drivers/thermal/tegra3_thermal.c line 209

Quoted 13 lines
drivers/thermal/tegra3_thermal.c new file +13 -0
+	if (!pdata) {
+		dev_err(&pdev->dev, "Get platform data failed.\n");
+		return -EINVAL;
+	}
+
+	/* Create a thermal zone */
+	tz = create_thermal_zone("tz_tegra", NULL);
+	if (!tz) {
+		dev_err(&pdev->dev, "Create thermal_zone failed.\n");
+		return -EINVAL;
+	}
+
+	pdata->tz = tz;

This isn't how you deal with driver data. Set driver data against a platform device using platform_set_drvdata(pdev, tz).

Yes, I didn't consider it carefully.
As Stephen said, our tegra will only support DT, so I will remove the platform date.

Thanks.
Wei.

Quoted 8 lines

drivers/thermal/tegra3_thermal.c line 255

Quoted 4 lines
drivers/thermal/tegra3_thermal.c new file +3 -0
+static int tegra30_thermal_remove(struct platform_device *pdev)
+{
+	struct tegra_thermal_data *pdata = pdev->dev.platform_data;

and use platform_get_drvdata() here - and don't use pdata->tz.

	struct struct thermal_zone *tz = platform_get_drvdata(pdev);
author
#38 lore raw

RFC PATCH 8/9 ARM: dt: t30 cardhu: add dt entry for thermal driver

In reply to Wei Ni

Enable thermal driver in the dts file.
Set sensor as lm90 remote sensor, and set throttle data.

---
 .../devicetree/bindings/thermal/tegra3-thermal.txt |   41 ++++++++++++++++++++
 arch/arm/boot/dts/tegra30-cardhu.dtsi              |   19 +++++++++
 2 files changed, 60 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/thermal/tegra3-thermal.txt
Documentation/devicetree/bindings/thermal/tegra3-thermal.txt new file +41 -0
new file mode 100644
index 0000000..dc3f922
@@ -0,0 +1,41 @@
+* Nvidia Tegra30 Thermal
+
+** Thermal node properties:
+
+- compatible : "nvidia,tegra30-thermal";
+- sensors: the sensor device node which we want to use in the thermal zone,
+  the arguments is the index of the sensor in sensor device node;
+- passive-delay: passive delay;
+- num-passive-trips : number of passive trip points, this is required, set
+  it 0 if none, if greater than 0, the following properties must be defined;
+- passive-trips : temperature of passive trip points;
+- num-active-trips: number of active trip points, this is required, set
+  it 0 if none, if greater than 0, the following properties must be defined;
+- active-trips: temperature of active trip points;
+- throt-tab-size: size of the throttle table, it's the max cooling state.
+- throt-tab: throttle table. the cooling state will be defined according to
+  this table.
+
+Usually these properties are separated in board related dts files.
+
+Example:
+thermal {
+	compatible = "nvidia,tegra30-thermal";
+	sensors = <&nct1008 0>;
+	passive-delay = <2000>;
+	num-passive-trips = <3>;
+	passive-trips = <70 80 90>;
+	num-active-trips = <4>;
+	active-trips = < 60 70 80 90>;
+	throt-tab-size = <10>;
+	throt-tab = <0 1000
+		640000 1000
+		640000 1000
+		640000 1000
+		640000 1000
+		640000 1000
+		760000 1050
+		760000 1050
+		1000000 1050
+		1000000 1050>;
+};
arch/arm/boot/dts/tegra30-cardhu.dtsi +19 -0
index 3f6ab89..9748b9b 100644
@@ -492,4 +492,23 @@
 		nvidia,spkr-en-gpios = <&wm8903 2 0>;
 		nvidia,hp-det-gpios = <&gpio 178 0>; /* gpio PW2 */
 	};
+
+	thermal {
+		compatible = "nvidia,tegra30-thermal";
+		sensors = <&nct1008 0>;
+		passive-delay = <2000>;
+		num-passive-trips = <1>;
+		passive-trips = <80>;
+		throt-tab-size = <10>;
+		throt-tab = <0 1000
+			640000 1000
+			640000 1000
+			640000 1000
+			640000 1000
+			640000 1000
+			760000 1050
+			760000 1050
+			1000000 1050
+			1000000 1050>;
+	};
 };
Signature · 1 line
1.7.9.5
#39 lore raw

RFC PATCH 8/9 ARM: dt: t30 cardhu: add dt entry for thermal driver

Documentation/devicetree/bindings/thermal/tegra3-thermal.txt

Quoted 2 lines · Wei Ni
Documentation/devicetree/bindings/thermal/tegra3-thermal.txt new file +0 -0
new file mode 100644

This should go with the previous patch (which introduces the driver) instead of this one. Bindings should be documented with their driver.

Alex.

author
#40 lore raw

RFC PATCH 8/9 ARM: dt: t30 cardhu: add dt entry for thermal driver

Quoted 6 lines · Alex Courbot

Documentation/devicetree/bindings/thermal/tegra3-thermal.txt

Quoted 2 lines · Wei Ni
Documentation/devicetree/bindings/thermal/tegra3-thermal.txt new file +0 -0
new file mode 100644

This should go with the previous patch (which introduces the driver) instead of this one. Bindings should be documented with their driver.

Ok, I will change it.

Wei.

Quoted 3 lines

Alex.

#41 lore raw

RFC PATCH 8/9 ARM: dt: t30 cardhu: add dt entry for thermal driver

In reply to Wei Ni

Documentation/devicetree/bindings/thermal/tegra3-thermal.txt

Quoted 4 lines · Wei Ni

Enable thermal driver in the dts file.
Set sensor as lm90 remote sensor, and set throttle data.

Documentation/devicetree/bindings/thermal/tegra3-thermal.txt +0 -0

The DT documentation should be either a separate patch before the one which implements it in the driver, or part of the driver patch. It shouldn't be part of the patch which instantiates the binding in the board DT file.

Quoted 1 line

+* Nvidia Tegra30 Thermal

NVIDIA is all capitals.

It seems like a word is missing: Thermal Management? Control/Zone?

Documentation/devicetree/bindings/thermal/tegra3-thermal.txt line 7

Quoted 5 lines
Documentation/devicetree/bindings/thermal/tegra3-thermal.txt +5 -0
+** Thermal node properties:
+
+- compatible : "nvidia,tegra30-thermal";
+- sensors: the sensor device node which we want to use in the thermal zone,
+  the arguments is the index of the sensor in sensor device node;

The arguments should be defined by the DT binding documentation for that referenced device, not by the referencing device.

DT properties that are custom defined for this binding should include the vendor prefix in their name. So, that'd be "nvidia,sensors". This may apply less if a "sensors" becomes a more generic standard, but it surely applies to all the other properties below.

Quoted 1 line

+- passive-delay: passive delay;

What does this mean?

Quoted 3 lines
+- num-passive-trips : number of passive trip points, this is required, set
+  it 0 if none, if greater than 0, the following properties must be defined;
+- passive-trips : temperature of passive trip points;

Presumably "num-passive-trips" can be calculated simply by counting the number of entries in "passive-trips"?

Please describe what passive and active trips are; someone who just reads this binding document (perhaps in conjunction with some HW documentation and/or other binding documentation) should be able to fill in this binding without other knowledge.

Quoted 3 lines
+- num-active-trips: number of active trip points, this is required, set
+  it 0 if none, if greater than 0, the following properties must be defined;
+- active-trips: temperature of active trip points;

Both comments above apply here too.

Quoted 3 lines
+- throt-tab-size: size of the throttle table, it's the max cooling state.
+- throt-tab: throttle table. the cooling state will be defined according to
+  this table.

Both comments above apply here too.

Also, what units are used for all these properties?

Judging by the example below, this property is a list of tuples. The meaning of each field in the tuple needs to be explained.

What happens when the CPU/SoC needs to be throttled? Must some clock or voltage be lowered/limited? If so, you need properties that indicate which clock/voltage/... needs to be acted upon.

Quoted 1 line

+Usually these properties are separated in board related dts files.

That's not really relevant.

author
#42 lore raw

RFC PATCH 8/9 ARM: dt: t30 cardhu: add dt entry for thermal driver

Quoted 10 lines · Stephen Warren

Documentation/devicetree/bindings/thermal/tegra3-thermal.txt

Quoted 4 lines · Wei Ni

Enable thermal driver in the dts file.
Set sensor as lm90 remote sensor, and set throttle data.

Documentation/devicetree/bindings/thermal/tegra3-thermal.txt +0 -0

The DT documentation should be either a separate patch before the one which implements it in the driver, or part of the driver patch. It shouldn't be part of the patch which instantiates the binding in the board DT file.

Ok, I will do it.

Quoted 4 lines
Quoted 2 lines

+* Nvidia Tegra30 Thermal

NVIDIA is all capitals.

Got it.

Quoted 2 lines

It seems like a word is missing: Thermal Management? Control/Zone?

It's better to use "NVIDIA Tegra30 Thermal Zone".

Quoted 10 lines

Documentation/devicetree/bindings/thermal/tegra3-thermal.txt line 7

Quoted 6 lines
Documentation/devicetree/bindings/thermal/tegra3-thermal.txt +5 -0
+** Thermal node properties:
+
+- compatible : "nvidia,tegra30-thermal";
+- sensors: the sensor device node which we want to use in the thermal zone,
+  the arguments is the index of the sensor in sensor device node;

The arguments should be defined by the DT binding documentation for that referenced device, not by the referencing device.

Quoted 4 lines

DT properties that are custom defined for this binding should include the vendor prefix in their name. So, that'd be "nvidia,sensors". This may apply less if a "sensors" becomes a more generic standard, but it surely applies to all the other properties below.

So these properties should be introduced in the DT binding documentation of the thermal framework.

Quoted 4 lines
Quoted 2 lines

+- passive-delay: passive delay;

What does this mean?

This is used for the thermal framework. As you said, this should be introduced with the framework.

Quoted 7 lines
Quoted 4 lines
+- num-passive-trips : number of passive trip points, this is required, set
+  it 0 if none, if greater than 0, the following properties must be defined;
+- passive-trips : temperature of passive trip points;

Presumably "num-passive-trips" can be calculated simply by counting the number of entries in "passive-trips"?

Oh, yes, I can remove it.

Quoted 5 lines

Please describe what passive and active trips are; someone who just reads this binding document (perhaps in conjunction with some HW documentation and/or other binding documentation) should be able to fill in this binding without other knowledge.

Durgadoss's patches introduced these value in the framework documentation, but he didn't touch the DT. So I will add it.

Quoted 7 lines
Quoted 4 lines
+- num-active-trips: number of active trip points, this is required, set
+  it 0 if none, if greater than 0, the following properties must be defined;
+- active-trips: temperature of active trip points;

Both comments above apply here too.

Quoted 7 lines
Quoted 3 lines
+- throt-tab-size: size of the throttle table, it's the max cooling state.
+- throt-tab: throttle table. the cooling state will be defined according to
+  this table.

Both comments above apply here too.

Also, what units are used for all these properties?

they use "int".
And according to our downstream dvfs driver, the throttle table use "unsigned long".

Quoted 7 lines

Judging by the example below, this property is a list of tuples. The meaning of each field in the tuple needs to be explained.

What happens when the CPU/SoC needs to be throttled? Must some clock or voltage be lowered/limited? If so, you need properties that indicate which clock/voltage/... needs to be acted upon.

This table which will be used by our DVFS driver, although it didn't be upstreamed yet. The table set clock frequency limited value with different cooling state. When the temperature touch the different limited value, we will set difficult cooling state, find the limited freq from this table and pass to the dvfs driver.
I think may be this table should be set for cooling device node. This table is only for our tegra dvfs, so I think we can parse this table in the tegra3_cooling.c, which will be the new driver for cooling device.

Quoted 4 lines
Quoted 2 lines

+Usually these properties are separated in board related dts files.

That's not really relevant.

Ok.

Quoted 1 line
#43 lore raw

RFC PATCH 8/9 ARM: dt: t30 cardhu: add dt entry for thermal driver

Quoted 5 lines · Wei Ni
Quoted 4 lines · Stephen Warren
Quoted 2 lines · Wei Ni

Enable thermal driver in the dts file.
Set sensor as lm90 remote sensor, and set throttle data.

Quoted 7 lines
Quoted 3 lines

Documentation/devicetree/bindings/thermal/tegra3-thermal.txt

Quoted 1 line
Documentation/devicetree/bindings/thermal/tegra3-thermal.txt +0 -0

Also, what units are used for all these properties?

they use "int".
And according to our downstream dvfs driver, the throttle table use "unsigned long".

int and unsigned long are types, not units. Units means degrees C vs. degrees F vs. degrees K, or Hz vs. KHz vs. MHz, etc.

Quoted 7 lines
Quoted 6 lines

Judging by the example below, this property is a list of tuples. The meaning of each field in the tuple needs to be explained.

What happens when the CPU/SoC needs to be throttled? Must some clock or voltage be lowered/limited? If so, you need properties that indicate which clock/voltage/... needs to be acted upon.

Quoted 8 lines

This table which will be used by our DVFS driver, although it didn't be upstreamed yet. The table set clock frequency limited value with different cooling state. When the temperature touch the different limited value, we will set difficult cooling state, find the limited freq from this table and pass to the dvfs driver.
I think may be this table should be set for cooling device node. This table is only for our tegra dvfs, so I think we can parse this table in the tegra3_cooling.c, which will be the new driver for cooling device.

Note that I expect the answers to these questions (pretty much all the questions I asked in response to your patch) to be included in the DT binding documentation, not just in email.

author
#44 lore raw

RFC PATCH 9/9 ARM: tegra: defconfig: enable thermal framework

In reply to Wei Ni

Enable tegra30 thermal framework.
New options enabled:
* TEGRA30_THERMAL: for tegra30 thermal framework.
* THERMAL: dependency for TEGRA30_THERMAL.

---
 arch/arm/configs/tegra_defconfig |    2 ++
 1 file changed, 2 insertions(+)
arch/arm/configs/tegra_defconfig +2 -0
index a7827fd..4aa3356 100644
@@ -128,6 +128,8 @@ CONFIG_GPIO_TPS65910=y
 CONFIG_POWER_SUPPLY=y
 CONFIG_BATTERY_SBS=y
 CONFIG_SENSORS_LM90=y
+CONFIG_THERMAL=y
+CONFIG_TEGRA30_THERMAL=y
 CONFIG_MFD_TPS6586X=y
 CONFIG_MFD_TPS65910=y
 CONFIG_MFD_MAX8907=y
Signature · 1 line
1.7.9.5