Re: [PATCH RFC 2/2] mfd: add Renesas RPC-IF driver

[Sergei Shtylyov <sergei.shtylyov@xxxxxxxxxxxxxxxxxx>]

On 06/19/2019 10:54 AM, Lee Jones wrote:

>>>>> Looks like a flash device to me.
>>>>
>>>>    More like a multi-protocol flash controller, with the real flash 
>>>> chip connected to it.
>>>
>>> Right, this has been my point from the start.
>>>
>>> It's a flash controller.  Sure, you can access it in different ways,

   No, the software access will be the same, just the initial controler
setup will be somewhat different depending on the flash type used...

>>> but it's still *just* a flash controller and thus not a true MFD.

   Also a SPI controller when a SPI bus is used.

>>> Surely this whole thing, including the shared portion should live in
>>> one of the memory related subsystems?
>>>
>>> This is not the first device people have tried to shove in MFD, that
>>> is really *just* an <X> device, able to be controlled via different
>>> protocols.

   You somehow still mix the (master) controller and (slave) device,
it seems...

>>> MFD is for registering child devices of chips which offer genuine
>>> cross-subsystem functionality.  It is not designed for mode selecting,
>>> or as a place to shove shared code just because a better location
>>> doesn't appear to exist.

   OK, fair enough...

>>> Also, ramming it into drivers/platform/<vendor> is not correct either,
>>> since this is not a platform controller driver either.

>> I will patch RPC-IF back to SPI only and 
>> rebase onto previous patches as bellow:
> 
> This sounds more like the easy way out, rather than the right thing to
> do.  Just because this isn't an MFD, doesn't mean it's not suitable
> for inclusion into the kernel.  Take a look at drivers/memory/Kconfig,
> and see if any of those devices sound familiar.

   TI AEMIF sounded familiar, I have some DaVinci/DA8xx background.
Trying to wrap my head into (missing?) API in drivers/memory/...

MBR, Sergei