[PATCH] hwmon: HIH-6130: Support I2C bus drivers without I2C_FUNC_SMBUS_QUICK

[PATCH] hwmon: HIH-6130: Support I2C bus drivers without I2C_FUNC_SMBUS_QUICK

[José Miguel Gonçalves]

Some I2C bus drivers do not allow zero-lenght data transfers which are
required to start a mesurement with the HIH6130/1 sensor. Nevertheless,
we can overcome this limitation by writing a zero dummy byte. This byte
is ignored by the sensor and was verified to be working with the OMAP
I2C bus driver in a BeagleBone board.

Signed-off-by: José Miguel Gonçalves <jose.goncalves@inov.pt>
---
 drivers/hwmon/hih6130.c |   16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/drivers/hwmon/hih6130.c b/drivers/hwmon/hih6130.c
index 2dc37c7..a860cc4 100644
--- a/drivers/hwmon/hih6130.c
+++ b/drivers/hwmon/hih6130.c
@@ -121,8 +121,20 @@ static int hih6130_update_measurements(struct i2c_client *client)
 	 */
 	if (time_after(jiffies, hih6130->last_update + HZ) || !hih6130->valid) {
 
-		/* write to slave address, no data, to request a measurement */
-		ret = i2c_master_send(client, tmp, 0);
+		/*
+		 * Write to slave address, to request a measurement.
+		 * According with the datasheet it should be with no data, but
+		 * for systems with I2C bus drivers that do not allow zero
+		 * length packets we write one dummy byte to allow sensor
+		 * measurements on them.
+		 */
+		if (i2c_get_functionality(client->adapter) &
+		    I2C_FUNC_SMBUS_QUICK) {
+			ret = i2c_master_send(client, tmp, 0);
+		} else {
+			tmp[0] = 0;
+			ret = i2c_master_send(client, tmp, 1);
+		}
 		if (ret < 0)
 			goto out;
 
-- 
1.7.10.4

[PATCH] hwmon: HIH-6130: Support I2C bus drivers without I2C_FUNC_SMBUS_QUICK

[José Miguel Gonçalves]

Some I2C bus drivers do not allow zero-length data transfers which are
required to start a measurement with the HIH6130/1 sensor. Nevertheless,
we can overcome this limitation by writing a zero dummy byte. This byte
is ignored by the sensor and was verified to be working with the OMAP
I2C bus driver in a BeagleBone board.

Signed-off-by: José Miguel Gonçalves <jose.goncalves@inov.pt>
---
v2: Use a field in struct hih6130 to store the measurement request's 
    write length when probing.

 drivers/hwmon/hih6130.c |   16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/drivers/hwmon/hih6130.c b/drivers/hwmon/hih6130.c
index 2dc37c7..8e46265 100644
--- a/drivers/hwmon/hih6130.c
+++ b/drivers/hwmon/hih6130.c
@@ -43,6 +43,7 @@
  * @last_update: time of last update (jiffies)
  * @temperature: cached temperature measurement value
  * @humidity: cached humidity measurement value
+ * @write_length: length for I2C measurement request
  */
 struct hih6130 {
 	struct device *hwmon_dev;
@@ -51,6 +52,7 @@ struct hih6130 {
 	unsigned long last_update;
 	int temperature;
 	int humidity;
+	size_t write_length;
 };
 
 /**
@@ -121,8 +123,15 @@ static int hih6130_update_measurements(struct i2c_client *client)
 	 */
 	if (time_after(jiffies, hih6130->last_update + HZ) || !hih6130->valid) {
 
-		/* write to slave address, no data, to request a measurement */
-		ret = i2c_master_send(client, tmp, 0);
+		/*
+		 * Write to slave address to request a measurement.
+		 * According with the datasheet it should be with no data, but
+		 * for systems with I2C bus drivers that do not allow zero
+		 * length packets we write one dummy byte to allow sensor
+		 * measurements on them.
+		 */
+		tmp[0] = 0;
+		ret = i2c_master_send(client, tmp, hih6130->write_length);
 		if (ret < 0)
 			goto out;
 
@@ -252,6 +261,9 @@ static int hih6130_probe(struct i2c_client *client,
 		goto fail_remove_sysfs;
 	}
 
+	hih6130->write_length = (i2c_get_functionality(client->adapter) &
+				 I2C_FUNC_SMBUS_QUICK ? 0 : 1);
+
 	return 0;
 
 fail_remove_sysfs:
-- 
1.7.10.4

Re: [PATCH] hwmon: HIH-6130: Support I2C bus drivers without I2C_FUNC_SMBUS_QUICK

[Guenter Roeck]

On Wed, Dec 11, 2013 at 11:11:13AM +0000, José Miguel Gonçalves wrote:
> Some I2C bus drivers do not allow zero-length data transfers which are
> required to start a measurement with the HIH6130/1 sensor. Nevertheless,
> we can overcome this limitation by writing a zero dummy byte. This byte
> is ignored by the sensor and was verified to be working with the OMAP
> I2C bus driver in a BeagleBone board.
> 
> Signed-off-by: José Miguel Gonçalves <jose.goncalves@inov.pt>

Applied, after simplifying write_length initialization a bit.

Should this go to -stable ?

Guenter

Re: [PATCH] hwmon: HIH-6130: Support I2C bus drivers without I2C_FUNC_SMBUS_QUICK

[José Miguel Gonçalves]

On 11-12-2013 19:29, Guenter Roeck wrote:
> On Wed, Dec 11, 2013 at 11:11:13AM +0000, José Miguel Gonçalves wrote:
>> Some I2C bus drivers do not allow zero-length data transfers which are
>> required to start a measurement with the HIH6130/1 sensor. Nevertheless,
>> we can overcome this limitation by writing a zero dummy byte. This byte
>> is ignored by the sensor and was verified to be working with the OMAP
>> I2C bus driver in a BeagleBone board.
>>
>> Signed-off-by: José Miguel Gonçalves <jose.goncalves@inov.pt>
> Applied, after simplifying write_length initialization a bit.
>
> Should this go to -stable ?

Any BeagleBone user that tries to use this sensor will encounter this issue. As 
this board is very popular, this will potentially solve the problems of many 
users. Nevertheless, it does not seems to me that this patch fixes "something 
critical", which is one of the conditions to be accepted in the -stable tree.

José Gonçalves

Re: [PATCH] hwmon: HIH-6130: Support I2C bus drivers without I2C_FUNC_SMBUS_QUICK

[Guenter Roeck]

On Wed, Dec 11, 2013 at 07:51:54PM +0000, José Miguel Gonçalves wrote:
> On 11-12-2013 19:29, Guenter Roeck wrote:
> >On Wed, Dec 11, 2013 at 11:11:13AM +0000, José Miguel Gonçalves wrote:
> >>Some I2C bus drivers do not allow zero-length data transfers which are
> >>required to start a measurement with the HIH6130/1 sensor. Nevertheless,
> >>we can overcome this limitation by writing a zero dummy byte. This byte
> >>is ignored by the sensor and was verified to be working with the OMAP
> >>I2C bus driver in a BeagleBone board.
> >>
> >>Signed-off-by: José Miguel Gonçalves <jose.goncalves@inov.pt>
> >Applied, after simplifying write_length initialization a bit.
> >
> >Should this go to -stable ?
> 
> Any BeagleBone user that tries to use this sensor will encounter
> this issue. As this board is very popular, this will potentially
> solve the problems of many users. Nevertheless, it does not seems to
> me that this patch fixes "something critical", which is one of the
> conditions to be accepted in the -stable tree.
> 
Depends on the definition of "critical".

Guenter