Re: [PATCH v3 3/5] mfd: tps65912: Add driver for the TPS65912 PMIC

["Andrew F. Davis" <afd@ti.com>]

On 09/29/2015 02:19 AM, Lee Jones wrote:
> On Mon, 28 Sep 2015, Andrew F. Davis wrote:
>
>> On 09/25/2015 04:24 PM, Lee Jones wrote:
>>> On Fri, 25 Sep 2015, Andrew F. Davis wrote:
>>>
>>>> On 09/25/2015 11:50 AM, Lee Jones wrote:
>>>>> On Thu, 24 Sep 2015, Andrew F. Davis wrote:
>>>>>
>>>>>> This patch adds support for TPS65912 mfd device. It provides
>>>>>> communication through the I2C and SPI interfaces. It contains
>>>>>> the following components:
>>>>>>
>>>>>>   - Regulators
>>>>>>   - GPIO controller
>>>>>>
>>>>>> Signed-off-by: Andrew F. Davis <afd@ti.com>
>>>>>> ---
>>>>>>   drivers/mfd/Kconfig          |  24 +++
>>>>>>   drivers/mfd/Makefile         |   3 +
>>>>>>   drivers/mfd/tps65912-core.c  | 114 +++++++++++++
>>>>>>   drivers/mfd/tps65912-i2c.c   |  86 ++++++++++
>>>>>>   drivers/mfd/tps65912-spi.c   |  85 ++++++++++
>>>>>>   include/linux/mfd/tps65912.h | 393 +++++++++++++++++++++++++++++++++++++++++++
>>>>>>   6 files changed, 705 insertions(+)
>>>>>>   create mode 100644 drivers/mfd/tps65912-core.c
>>>>>>   create mode 100644 drivers/mfd/tps65912-i2c.c
>>>>>>   create mode 100644 drivers/mfd/tps65912-spi.c
>>>>>>   create mode 100644 include/linux/mfd/tps65912.h
>>>>>
>>>>> [...]
>>>>>
>>>>>> +#define TPS65912_IRQ(_name, _reg, _offset)			\
>>>>>> +	[TPS65912_IRQ_ ## _name] = {				\
>>>>>> +		.mask = TPS65912_ ## _reg ## _ ## _name,	\
>>>>>> +		.reg_offset = _offset,				\
>>>>>> +	}
>>>>>
>>>>> I told you about this already.
>>>>>
>>>>> If you want this set to be merged for v4.3 then you'll need commit
>>>>> b4fe8ba ("regmap: Add generic macro to define regmap_irq") from my
>>>>> tree.
>>>>
>>>> You asked me to submit this to Mark Brown, I didn't realize you also
>>>> wanted me to use the one in your tree. Using yours will make my lines
>>>> over 80 chars and so it kind of defeats the purpose but I'll do it
>>>> for v4.
>>>
>>> I asked two of you to upstream a generic incarnation of your
>>> implementation.  The other guy got there first.  Now I want everyone
>>> to use it instead of hand-rolling their own.
>>>
>>
>> Got it.
>>
>>>>> [...]
>>>>>
>>>>>> +static struct i2c_driver tps65912_i2c_driver = {
>>>>>> +	.driver		= {
>>>>>> +		.name	= "tps65912",
>>>>>> +		.of_match_table = tps65912_i2c_of_match_table,
>>>>>
>>>>> of_match_ptr()
>>>>>
>>>>
>>>> Why? tps65912_i2c_of_match_table is always compiled in.
>>>
>>> But it needn't be.
>>>
>>
>> This driver is DT only, would it make more sense to just add a dependency
>> on CONFIG_OF in the Kconfig?
>
> Yes, that is the alternitive.
>
>> I don't see many other drivers that are DT
>> only doing this, the driver should compile just fine without OF enabled,
>> it just wont have DT functionality, so is it really dependent on OF?
>
> Then it should be "depends on OF || COMPILE_TEST"
>

OK, I'll add this.

>> The of_match_ptr(x) macro looks to be useful when the x is conditionally
>> compiled in, but in my case the x (my of_device_id table) is always compiled.
>