iio: mma8452: Fix probe failing when an i2c_device_id is used
PATCH 1/1 iio: mma8452: Fix probe failing when an i2c_device_id is used
The mma8452_driver declares both of_match_table and i2c_driver.id_table match-tables, but its probe() function only checked for of matches.
Add support for i2c_device_id matches. This fixes the driver not loading on some x86 tablets (e.g. the Nextbook Ares 8) where the i2c_client is instantiated by platform code using an i2c_device_id.
--- drivers/iio/accel/mma8452.c | 23 +++++++++++++++-------- 1 file changed, 15 insertions(+), 8 deletions(-)
drivers/iio/accel/mma8452.c +15 -8
@@ -1523,12 +1523,7 @@ static int mma8452_probe(struct i2c_client *client,struct iio_dev *indio_dev;int ret;const struct of_device_id *match; - - match =of_match_device (mma8452_dt_ids, &client->dev); -if (!match) { -dev_err (&client->dev,"unknown device model\n" ); -return -ENODEV ; - } +const char *compatible; indio_dev =devm_iio_device_alloc (&client->dev,sizeof (*data));if (!indio_dev)
@@ -1537,7 +1532,19 @@ static int mma8452_probe(struct i2c_client *client, data =iio_priv (indio_dev); data->client = client;mutex_init (&data->lock); - data->chip_info = match->data; + +if (id) { + compatible = id->name; + data->chip_info = &mma_chip_info_table[id->driver_data]; + }else { + match =of_match_device (mma8452_dt_ids, &client->dev); +if (!match) { +dev_err (&client->dev,"unknown device model\n" ); +return -ENODEV ; + } + compatible = match->compatible; + data->chip_info = match->data; + } data->vdd_reg =devm_regulator_get (&client->dev,"vdd" );if (IS_ERR (data->vdd_reg))
@@ -1581,7 +1588,7 @@ static int mma8452_probe(struct i2c_client *client, }dev_info (&client->dev,"registering %s accelerometer; ID 0x%x\n" , - match->compatible, data->chip_info->chip_id); + compatible, data->chip_info->chip_id);i2c_set_clientdata (client, indio_dev); indio_dev->info = &mma8452_info;
Signature · 1 line
2.33.1
Jonathan Cameron
maintainerQuoted 8 lines · On Thu, 6 Jan 2022 12:14:14 +0100 Hans de Goede <[email protected]> wrote:
The
mma8452_driverdeclares bothof_match_tableand i2c_driver.id_table match-tables, but itsprobe()function only checked for of matches.Add support for
i2c_device_idmatches. This fixes the driver not loading on some x86 tablets (e.g. the Nextbook Ares 8) where thei2c_clientis instantiated by platform code using ani2c_device_id.Signed-off-by Hans de Goede <[email protected]>
Hi Hans,
At some point we'll want to get rid of the of_ specific stuff in here in favour of generic firmware properties and I suspect at that time we'll move the device name into the chip_info_table[] entries so that we can just use device_get_match_data()
In the meantime this fix looks good to me. Is there an appropriate
Fixes: tag?
Thanks,
Jonathan
Quoted 52 lines
--- drivers/iio/accel/mma8452.c | 23 +++++++++++++++-------- 1 file changed, 15 insertions(+), 8 deletions(-)drivers/iio/accel/mma8452.c +15
-8@@ -1523,12 +1523,7 @@ static int mma8452_probe(struct i2c_client *client,struct iio_dev *indio_dev;int ret;const struct of_device_id *match; - - match =of_match_device (mma8452_dt_ids, &client->dev); -if (!match) { -dev_err (&client->dev,"unknown device model\n" ); -return -ENODEV ; - } +const char *compatible; indio_dev =devm_iio_device_alloc (&client->dev,sizeof (*data));if (!indio_dev)@@ -1537,7 +1532,19 @@ static int mma8452_probe(struct i2c_client *client, data =iio_priv (indio_dev); data->client = client;mutex_init (&data->lock); - data->chip_info = match->data; + +if (id) { + compatible = id->name; + data->chip_info = &mma_chip_info_table[id->driver_data]; + }else { + match =of_match_device (mma8452_dt_ids, &client->dev); +if (!match) { +dev_err (&client->dev,"unknown device model\n" ); +return -ENODEV ; + } + compatible = match->compatible; + data->chip_info = match->data; + } data->vdd_reg =devm_regulator_get (&client->dev,"vdd" );if (IS_ERR (data->vdd_reg))@@ -1581,7 +1588,7 @@ static int mma8452_probe(struct i2c_client *client, }dev_info (&client->dev,"registering %s accelerometer; ID 0x%x\n" , - match->compatible, data->chip_info->chip_id); + compatible, data->chip_info->chip_id);i2c_set_clientdata (client, indio_dev); indio_dev->info = &mma8452_info;
Hi,
Quoted 13 lines · Jonathan Cameron
Quoted 8 lines · On Thu, 6 Jan 2022 12:14:14 +0100 Hans de Goede <[email protected]> wrote:
The
mma8452_driverdeclares bothof_match_tableand i2c_driver.id_table match-tables, but itsprobe()function only checked for of matches.Add support for
i2c_device_idmatches. This fixes the driver not loading on some x86 tablets (e.g. the Nextbook Ares 8) where thei2c_clientis instantiated by platform code using ani2c_device_id.Signed-off-by Hans de Goede <[email protected]>Hi Hans,
Quoted 7 lines
At some point we'll want to get rid of the
of_specific stuff in here in favour of generic firmware properties and I suspect at that time we'll move the device name into the chip_info_table[] entries so that we can just usedevice_get_match_data()In the meantime this fix looks good to me. Is there an appropriate
Fixes: tag?
I did a quick dive in the git history and the of_match_device() || return -ENODEV behavior was introduced in:
c3cdd6e48e35 ("iio: mma8452: refactor for seperating chip specific data")
Regards,
Hans
Quoted 53 lines
Quoted 52 lines
--- drivers/iio/accel/mma8452.c | 23 +++++++++++++++-------- 1 file changed, 15 insertions(+), 8 deletions(-)drivers/iio/accel/mma8452.c +15
-8@@ -1523,12 +1523,7 @@ static int mma8452_probe(struct i2c_client *client,struct iio_dev *indio_dev;int ret;const struct of_device_id *match; - - match =of_match_device (mma8452_dt_ids, &client->dev); -if (!match) { -dev_err (&client->dev,"unknown device model\n" ); -return -ENODEV ; - } +const char *compatible; indio_dev =devm_iio_device_alloc (&client->dev,sizeof (*data));if (!indio_dev)@@ -1537,7 +1532,19 @@ static int mma8452_probe(struct i2c_client *client, data =iio_priv (indio_dev); data->client = client;mutex_init (&data->lock); - data->chip_info = match->data; + +if (id) { + compatible = id->name; + data->chip_info = &mma_chip_info_table[id->driver_data]; + }else { + match =of_match_device (mma8452_dt_ids, &client->dev); +if (!match) { +dev_err (&client->dev,"unknown device model\n" ); +return -ENODEV ; + } + compatible = match->compatible; + data->chip_info = match->data; + } data->vdd_reg =devm_regulator_get (&client->dev,"vdd" );if (IS_ERR (data->vdd_reg))@@ -1581,7 +1588,7 @@ static int mma8452_probe(struct i2c_client *client, }dev_info (&client->dev,"registering %s accelerometer; ID 0x%x\n" , - match->compatible, data->chip_info->chip_id); + compatible, data->chip_info->chip_id);i2c_set_clientdata (client, indio_dev); indio_dev->info = &mma8452_info;
Quoted 21 lines · On Sun, 9 Jan 2022 16:35:46 +0100 Hans de Goede <[email protected]> wrote:
Hi,
Quoted 18 lines · Jonathan Cameron
Quoted 8 lines · On Thu, 6 Jan 2022 12:14:14 +0100 Hans de Goede <[email protected]> wrote:
The
mma8452_driverdeclares bothof_match_tableand i2c_driver.id_table match-tables, but itsprobe()function only checked for of matches.Add support for
i2c_device_idmatches. This fixes the driver not loading on some x86 tablets (e.g. the Nextbook Ares 8) where thei2c_clientis instantiated by platform code using ani2c_device_id.Signed-off-by Hans de Goede <[email protected]>Hi Hans,
At some point we'll want to get rid of the
of_specific stuff in here in favour of generic firmware properties and I suspect at that time we'll move the device name into the chip_info_table[] entries so that we can just usedevice_get_match_data()
Quoted 7 lines
Quoted 2 lines
In the meantime this fix looks good to me. Is there an appropriate
Fixes: tag?I did a quick dive in the git history and the
of_match_device()|| return-ENODEVbehavior was introduced in:
c3cdd6e48e35("iio: mma8452: refactor for seperating chip specific data")
Applied to the fixes-togreg branch of iio.git with that as the fixes tag.
Hmm. Sending "Fixes: tag?" was a bad idea on my part as b4 picked it up as a fixes tag. Good think I was editing it anyway or I might not have noticed.
Thanks,
Jonathan
Quoted 62 lines
Regards,
Hans
Quoted 53 lines
Quoted 52 lines
--- drivers/iio/accel/mma8452.c | 23 +++++++++++++++-------- 1 file changed, 15 insertions(+), 8 deletions(-)drivers/iio/accel/mma8452.c +15
-8@@ -1523,12 +1523,7 @@ static int mma8452_probe(struct i2c_client *client,struct iio_dev *indio_dev;int ret;const struct of_device_id *match; - - match =of_match_device (mma8452_dt_ids, &client->dev); -if (!match) { -dev_err (&client->dev,"unknown device model\n" ); -return -ENODEV ; - } +const char *compatible; indio_dev =devm_iio_device_alloc (&client->dev,sizeof (*data));if (!indio_dev)@@ -1537,7 +1532,19 @@ static int mma8452_probe(struct i2c_client *client, data =iio_priv (indio_dev); data->client = client;mutex_init (&data->lock); - data->chip_info = match->data; + +if (id) { + compatible = id->name; + data->chip_info = &mma_chip_info_table[id->driver_data]; + }else { + match =of_match_device (mma8452_dt_ids, &client->dev); +if (!match) { +dev_err (&client->dev,"unknown device model\n" ); +return -ENODEV ; + } + compatible = match->compatible; + data->chip_info = match->data; + } data->vdd_reg =devm_regulator_get (&client->dev,"vdd" );if (IS_ERR (data->vdd_reg))@@ -1581,7 +1588,7 @@ static int mma8452_probe(struct i2c_client *client, }dev_info (&client->dev,"registering %s accelerometer; ID 0x%x\n" , - match->compatible, data->chip_info->chip_id); + compatible, data->chip_info->chip_id);i2c_set_clientdata (client, indio_dev); indio_dev->info = &mma8452_info;