On 14/12/2022 10:32, Jonathan Cameron wrote:
> On Tue, 13 Dec 2022 19:53:30 +0100
> Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxx> wrote:
>
>> On 13/12/2022 18:15, Han Xu wrote:
>>> Add this new device entry in the driver id table.
>>>
>>> Signed-off-by: Han Xu <han.xu@xxxxxxx>
>>>
>>> ---
>>> changes in v2
>>> - change chip info orders
>>> ---
>>> drivers/iio/accel/fxls8962af-core.c | 7 +++++++
>>> drivers/iio/accel/fxls8962af-i2c.c | 2 ++
>>> drivers/iio/accel/fxls8962af.h | 1 +
>>> 3 files changed, 10 insertions(+)
>>>
>>> diff --git a/drivers/iio/accel/fxls8962af-core.c b/drivers/iio/accel/fxls8962af-core.c
>>> index 98811e4e16bb..c3589c3084ee 100644
>>> --- a/drivers/iio/accel/fxls8962af-core.c
>>> +++ b/drivers/iio/accel/fxls8962af-core.c
>>> @@ -127,6 +127,7 @@
>>> #define FXLS8962AF_DEVICE_ID 0x62
>>> #define FXLS8964AF_DEVICE_ID 0x84
>>> #define FXLS8974CF_DEVICE_ID 0x86
>>> +#define FXLS8967AF_DEVICE_ID 0x87
>>>
>>> /* Raw temp channel offset */
>>> #define FXLS8962AF_TEMP_CENTER_VAL 25
>>> @@ -765,6 +766,12 @@ static const struct fxls8962af_chip_info fxls_chip_info_table[] = {
>>> .channels = fxls8962af_channels,
>>> .num_channels = ARRAY_SIZE(fxls8962af_channels),
>>> },
>>> + [fxls8967af] = {
>>> + .chip_id = FXLS8967AF_DEVICE_ID,
>>> + .name = "fxls8967af",
>>> + .channels = fxls8962af_channels,
>>> + .num_channels = ARRAY_SIZE(fxls8962af_channels),
>>> + },
>>> [fxls8974cf] = {
>>> .chip_id = FXLS8974CF_DEVICE_ID,
>>> .name = "fxls8974cf",
>>> diff --git a/drivers/iio/accel/fxls8962af-i2c.c b/drivers/iio/accel/fxls8962af-i2c.c
>>> index 17dd56756ff9..a8944b255a28 100644
>>> --- a/drivers/iio/accel/fxls8962af-i2c.c
>>> +++ b/drivers/iio/accel/fxls8962af-i2c.c
>>> @@ -30,6 +30,7 @@ static int fxls8962af_probe(struct i2c_client *client)
>>> static const struct i2c_device_id fxls8962af_id[] = {
>>> { "fxls8962af", fxls8962af },
>>> { "fxls8964af", fxls8964af },
>>> + { "fxls8967af", fxls8967af },
>>> { "fxls8974cf", fxls8974cf },
>>> {}
>>> };
>>> @@ -38,6 +39,7 @@ MODULE_DEVICE_TABLE(i2c, fxls8962af_id);
>>> static const struct of_device_id fxls8962af_of_match[] = {
>>> { .compatible = "nxp,fxls8962af" },
>>> { .compatible = "nxp,fxls8964af" },
>>> + { .compatible = "nxp,fxls8967af" },
>>> { .compatible = "nxp,fxls8974cf" },
>>
>> This is confusing. The I2C ID table has driver data, but OF ID table
>> hasn't. So are they compatible or not?
>
> Due to some evilness in i2c that 'works' as long as the two arrays have
> matching entries. As a general rule we prefer to have the data in both, check
> the firmware table first and only then fallback to i2c_device_id data on the
> basis it is less fragile.
>
> The evilness in i2c is that the search for match data will use the dt compatible
> stripped of the vendor prefix and string match that against the i2c_device_id table.
>
> Nice to clean this up, but not necessarily in this series (fine if it is though!)
OK, so in fact devices are not fully compatible - I got mislead by OF
table. I'll comment in bindings about it.
Best regards,
Krzysztof
