Re: [PATCH 1/3] spi: Revert "spi: Remove unused function spi_busnum_to_master()"

[Hans de Goede]

Hi,

On 12/3/21 12:06, Andy Shevchenko wrote:
> On Thu, Dec 02, 2021 at 04:24:19PM +0000, Stefan Binding wrote:
>> From: Lucas Tanure <tanureal@xxxxxxxxxxxxxxxxxxxxx>
>>
>> Revert commit bdc7ca008e1f ("spi: Remove unused function
>> spi_busnum_to_master()")
>> This function is needed for the spi version of i2c multi
>> instantiate driver.
> 
> I understand the intention, but I have no clue from this series (it lacks of
> proper cover letter, it lacks of much better and justified commit message in
> the patch 3) what is the actual issue. Without these to be provided it's no go
> for the series. Please, provide much better description what is the actual
> issue you are trying to solve (from patch 3 my guts telling me that this can
> be achieved differently without this code being involved).

Yes I assume that eventually there will be a follow-up which will
actually add some new ACPI HIDs to the new bus-multi-instantiate.c file ?

Can we please have (some of) those patches as part of the next
version, so that we can see how you will actually use this?

Also I'm wondering is this actually about ACPI device's having multiple
SpiSerialBusV2 resources in a single _CRS resource list ?

Or do you plan to use this for devices with only a single
I2cSerialBusV2 or SpiSerialBusV2 resource to e.g. share IRQ lookup
code between the 2 cases ?

If you plan to use this for devices with only a single
I2cSerialBusV2 or SpiSerialBusV2 resource, then I'm going to have to
nack this.

Each ACPI HID which needs to be handled in this code also needs an
entry in the i2c_multi_instantiate_ids[] list in drivers/acpi/scan.c
which is code which ends up loaded on every single ACPI system, so
we really only want to add HIDs there for the special case of having
multiple I2cSerialBusV2 or SpiSerialBusV2 resources in a single
ACPI Device / ACPI fwnode.

If you are looking to use this as a way to share code for other reasons
(which is a good goal to strive for!) please find another way, such
as e.g. adding some helper functions to drivers/gpio/gpiolib-acpi.c
(note there already are a couple of helpers there which you may use).

Regards,

Hans