Skip to contentSkip to conversation

hwmon: ADC124S501 generic driver

RFC PATCH Superseded Marc Pignat 2 participants 2 messages Jul 23, 2008 – Jul 29, 2008 Hardware monitoring

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

RFC PATCH 1/1 hwmon: ADC124S501 generic driver

SPI driver for analog to digital converters national semiconductor ADC081S101, ADC124S501, ...

Hi all!

patch against 2.6.26, tested on a custom arm board and only compil-tested on x86.

This driver add support for National Semiconductor ADC<bb><c>S<sss> chip family, where:
* bb is the resolution in number of bits (8, 10, 12)
* c is the number of channels (1, 2, 4)
* sss is the maximum conversion speed (021 for 200 kSPS, 051 for 500 kSPS and 101 for 1 MSPS)

Some remarks:
* The chip type (-> the number of inputs) are determined by the module alias, is it a good idea? it could be implemented using platform data.

* The Makefile seems ordered alphabetically, what order should be used for the Konfig file?

Best regards

Marc

drivers/hwmon/Kconfig +16 -0
index 00ff533..9c4a579 100644
@@ -390,6 +390,22 @@ config SENSORS_LM70
 	  This driver can also be built as a module.  If so, the module
 	  will be called lm70.
 
+config SENSORS_ADCXX
+	tristate "National Semiconductor ADCxxxSxxx"
+	depends on SPI_MASTER && EXPERIMENTAL
+	help
+	  If you say yes here you get support for the National Semiconductor
+	  ADC<bb><c>S<sss> chip family, where
+	  * bb  is the resolution in number of bits (8, 10, 12)
+	  * c   is the number of channels (1, 2, 4)
+	  * sss is the maximum conversion speed (021 for 200 kSPS, 051 for 500
+	    kSPS and 101 for 1 MSPS)
+
+	  Examples : ADC081S101, ADC124S501, ...
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called adcxx.
+
 config SENSORS_LM75
 	tristate "National Semiconductor LM75 and compatibles"
 	depends on I2C
drivers/hwmon/Makefile +1 -0
index d098677..026ba1e 100644
@@ -16,6 +16,7 @@ obj-$(CONFIG_SENSORS_W83791D)	+= w83791d.o
 obj-$(CONFIG_SENSORS_ABITUGURU)	+= abituguru.o
 obj-$(CONFIG_SENSORS_ABITUGURU3)+= abituguru3.o
 obj-$(CONFIG_SENSORS_AD7418)	+= ad7418.o
+obj-$(CONFIG_SENSORS_ADCXX)	+= adcxx.o
 obj-$(CONFIG_SENSORS_ADM1021)	+= adm1021.o
 obj-$(CONFIG_SENSORS_ADM1025)	+= adm1025.o
 obj-$(CONFIG_SENSORS_ADM1026)	+= adm1026.o
drivers/hwmon/adcxx.c new file +292 -0

292 lines not on this page · raw message

_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

#1 lore raw
Quoted 17 lines · On Wed, 23 Jul 2008 13:07:54 +0200 Marc Pignat <[email protected]> wrote:

SPI driver for analog to digital converters national semiconductor ADC081S101, ADC124S501, ...

Hi all!

patch against 2.6.26, tested on a custom arm board and only compil-tested on x86.

This driver add support for National Semiconductor ADC<bb><c>S<sss> chip family, where:
* bb is the resolution in number of bits (8, 10, 12)
* c is the number of channels (1, 2, 4)
* sss is the maximum conversion speed (021 for 200 kSPS, 051 for 500 kSPS and 101 for 1 MSPS)

Quoted 7 lines

Some remarks:
* The chip type (-> the number of inputs) are determined by the module alias, is it a good idea? it could be implemented using platform data.

* The Makefile seems ordered alphabetically, what order should be used for the Konfig file?

That changelog is a bit of a mess. I cleaned up up as:

SPI driver for analog to digital converters national semiconductor ADC081S101, ADC124S501, ...

This driver adds support for National Semiconductor ADC<bb><c>S<sss> chip family, where:

* bb is the resolution in number of bits (8, 10, 12)
* c is the number of channels (1, 2, 4)
* sss is the maximum conversion speed (021 for 200 kSPS, 051 for 500 kSPS and 101 for 1 MSPS)

Some remarks:

* The chip type (-> the number of inputs) are determined by the module alias, is it a good idea? it could be implemented using platform data.

Cc "Mark M. Hoffman" <[email protected]>
Cc Jean Delvare <[email protected]>
Cc David Brownell <[email protected]>

but then didn't apply the patch.

Quoted 5 lines

...

drivers/hwmon/adcxx.c new file +2 -0
+/* sysfs hook function */
+static ssize_t adcxx_read(struct device *dev,

drivers/hwmon/adcxx.c line 68

Quoted 10 lines
drivers/hwmon/adcxx.c new file +10 -0
+		struct device_attribute *devattr, char *buf)
+{
+	struct spi_device *spi = to_spi_device(dev);
+	struct sensor_device_attribute *attr = to_sensor_dev_attr(devattr);
+	struct adcxx *adc = dev_get_drvdata(&spi->dev);
+
+	u8 tx_buf[2] = { attr->index << 3 }; /* other bits are don't care */
+	u8 rx_buf[2];
+	int status;
+	int value;

The driver has rather a lot of inexplicable blank lines in the middle of the automatic variable definitions. The preferred style is no blank lines within the definitions and a single blank line after them all, thanks.

Quoted 36 lines

...

drivers/hwmon/adcxx.c new file +33 -0
+static int __devinit adcxx_probe(struct spi_device *spi, int channels)
+{
+	struct adcxx *adc;
+	int status;
+	int i;
+
+	adc = kzalloc(sizeof *adc, GFP_KERNEL);
+	if (!adc)
+		return -ENOMEM;
+
+	/* set a default value for the reference */
+	adc->reference = 3300;
+
+	adc->channels = channels;
+
+	mutex_init(&adc->lock);
+
+	dev_set_drvdata(&spi->dev, adc);
+
+	for (i = 0; i < 3 + adc->channels; i++) {
+		status = device_create_file(&spi->dev, &ad_input[i].dev_attr);
+		if (status)
+			goto out_dev_create_file_failed;
+	}
+
+	adc->hwmon_dev = hwmon_device_register(&spi->dev);
+	if (IS_ERR(adc->hwmon_dev)) {
+		dev_dbg(&spi->dev, "hwmon_device_register failed.\n");
+		status = PTR_ERR(adc->hwmon_dev);
+		goto out_dev_reg_failed;
+	}
+
+	return 0;

drivers/hwmon/adcxx.c line 204

Quoted 10 lines
drivers/hwmon/adcxx.c new file +10 -0
+
+out_dev_create_file_failed:
+	hwmon_device_unregister(adc->hwmon_dev);
+	for (i = 0; i < 3 + adc->channels; i++)
+		device_remove_file(&spi->dev, &ad_input[i].dev_attr);
+out_dev_reg_failed:
+	dev_set_drvdata(&spi->dev, NULL);
+	kfree(adc);
+	return status;
+}

The error recovery here is messed up. The targets of the `goto's are reversed. But even if that is fixed, we can end up doing device_remove_file() of objects which weren't successfully created. That might work, or it might generate runtime warnings or it might crash. I don't know. It'd be best to just avoid doing it?

_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors