Re: [PATCH v3 5/6] platform/x86: Add intel_skl_int3472 driver

[Daniel Scally]

Hi Hans, thanks for the input

On 22/02/2021 13:27, Hans de Goede wrote:
> Hi,
>
> On 2/22/21 2:19 PM, Daniel Scally wrote:
>> Hi all
>>
>> On 22/02/2021 13:07, Daniel Scally wrote:
>>> diff --git a/drivers/platform/x86/intel-int3472/Kconfig b/drivers/platform/x86/intel-int3472/Kconfig
>>> new file mode 100644
>>> index 000000000000..b94622245c21
>>> --- /dev/null
>>> +++ b/drivers/platform/x86/intel-int3472/Kconfig
>>> @@ -0,0 +1,31 @@
>>> +config INTEL_SKL_INT3472
>>> +	tristate "Intel SkyLake ACPI INT3472 Driver"
>>> +	depends on ACPI
>>> +	depends on REGULATOR
>>> +	depends on GPIOLIB
>>> +	depends on COMMON_CLK && CLKDEV_LOOKUP
>>> +	depends on I2C
>>> +	select MFD_CORE
>>> +	select REGMAP_I2C
>>> +	help
>>> +	  This driver adds support for the INT3472 ACPI devices found on some
>>> +	  Intel SkyLake devices.
>>> +
>>> +	  The INT3472 is an Intel camera power controller, a logical device
>>> +	  found on some Skylake-based systems that can map to different
>>> +	  hardware devices depending on the platform. On machines
>>> +	  designed for Chrome OS, it maps to a TPS68470 camera PMIC. On
>>> +	  machines designed for Windows, it maps to either a TP68470
>>> +	  camera PMIC, a uP6641Q sensor PMIC, or a set of discrete GPIOs
>>> +	  and power gates.
>>> +
>>> +	  If your device was designed for Chrome OS, this driver will provide
>>> +	  an ACPI OpRegion, which must be available before any of the devices
>>> +	  using it are probed. For this reason, you should select Y if your
>>> +	  device was designed for ChromeOS. For the same reason the
>>> +	  I2C_DESIGNWARE_PLATFORM option must be set to Y too.
>>> +
>>> +	  Say Y or M here if you have a SkyLake device designed for use
>>> +	  with Windows or ChromeOS. Say N here if you are not sure.
>>> +
>>> +	  The module will be named "intel-skl-int3472"
>> The Kconfig option for the existing tps68470 driver is a bool which
>> depends on I2C_DESIGNWARE_PLATFORM=y, giving the following reason:
>>
>> This option is a bool as it provides an ACPI operation
>> region, which must be available before any of the devices
>> using this are probed. This option also configures the
>> designware-i2c driver to be built-in, for the same reason.
>>
>> One problem I've faced is that that scenario only applies to some
>> devices that this new driver can support, so hard-coding it as built in
>> didn't make much sense. For that reason I opted to set it tristate, but
>> of course that issue still exists for ChromeOS devices where the
>> OpRegion will be registered. I opted for simply documenting that
>> requirement, as is done in aaac4a2eadaa6: "mfd: axp20x-i2c: Document
>> that this must be builtin on x86", but that's not entirely satisfactory.
>> Possible alternatives might be setting "depends on
>> I2C_DESIGNWARE_PLATFORM=y if CHROME_PLATFORMS" or something similar,
>> though of course the User would still have to realise they need to
>> build-in the INTEL_SKL_INT3472 Kconfig option too.
>>
>> Feedback around this issue would be particularly welcome, as I'm not
>> sure what the best approach might be.
> This is a tricky area, I actually wrote the "mfd: axp20x-i2c: Document
> that this must be builtin on x86" patch you refer to. At first I tried
> to express the dependency in Kconfig language but things got too complex
> and Kconfig sometimes became unhappy about circular deps (or something
> like that).


Yes, I had a go too; with similar results

> The most important thing here is to make sure that the generic configs
> shipped by distros get this right; and we can hope that people creating
> those configs at least read the help text...
>
> So all in all I believe that just documenting the requirement is fine.


OK - that's what I'm hoping is the consensus, as I don't think it can be
made _entirely_ seamless through dependencies or whatever anyway, in
which case documenting it seems like the cleanest approach to me.

>
> The alternative would be to just change I2C_DESIGNWARE_PLATFORM (and the
> core) to a bool, or at least make it not selectable as module when
> X86 and ACPI are set... That would be a bit of a big hammer but might
> not be the worst idea actually.
>
> Regards,
>
> Hans
>