On Wed, May 20, 2015 at 10:47 AM, Mika Westerberg wrote: > On Tue, May 19, 2015 at 05:03:29PM +0300, Robert Dolca wrote: >> For i2c devices enumerated with ACPI you need to declare both >> acpi_match_table and id_table. When using ACPI, the i2c_device_id structure >> supplied to the probe function is null and you have to handle this case >> in the driver. >> >> The current name for the i2c client when using ACPI is "HID:UID" where the >> UID has 7 or 8 characters and the UID has 2 characters. The UID is not >> relevant for identifying the chip so it does not have any practical >> purpose. > > First of all, it is not "HID:UID" since the number after ":" is actually > increasing number assigned by the ACPI core. Nothing to do with _UID. You are right. My mistake. > Secondly we do not list "_HID:nn" in drivers acpi_match_tables but > instead it is either "HID" or "CID", no ":nn" there. I didn't say that you do that :) > >> Modifying i2c_match_id we make the comparison by ignoring the UID from the >> client name when the device was discovered using ACPI. The comparison is >> case insensitive because the ACPI names are uppercase and the DT and ID >> table names are lowercase. It would not make sense to have two different >> chips with the same name and the only diference being the capitalized >> letters. >> >> With these changes the probe function gets a valid i2c_device_id and the >> driver doesn't have to declare acpi_match_table. > > No. We don't do that for DT and we definitely don't want to mix ACPI > identifiers with arbitrary I2C device names. > > You are not supposed to put ACPI identifiers into i2c_device_id table. Currently, if the name used for DT (in dts) matches one of the names specified in the id table you will have a match. Isn't that an intended behavior? Robert -- To unsubscribe from this list: send the line "unsubscribe linux-i2c" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
