Re: [PATCH 1/2] misc: add nirtfeatures driver

From: Lee Jones
Date: Wed Mar 16 2016 - 03:48:37 EST


On Mon, 14 Mar 2016, Josh Cartwright wrote:

> On Mon, Mar 14, 2016 at 03:05:59PM -0700, Greg KH wrote:
> > On Mon, Mar 14, 2016 at 04:54:32PM -0500, Kyle Roeschley wrote:
> > > From: Jeff Westfahl <jeff.westfahl@xxxxxx>
> > >
> > > This driver introduces support for hardware features of National
> > > Instruments real-time controllers. This is an ACPI device that exposes
> > > LEDs, switches, and watchdogs.
> >
> > If it's an acpi driver, why not put it in drivers/acpi?
>
> For the same reason we don't move all drivers for devices-on-a-PCI-bus
> into drivers/pci?
>
> Drivers typically exist in the sourcetree with other drivers which
> implement similar functionality, which works great for devices with
> clear functional boundaries (GPIO controller drivers in drivers/gpio,
> led drivers in drivers/leds, etc. etc.); but for devices which are a
> hodgepodge of functionality, there isn't really a good fit anywhere
> except maybe in misc or mfd.
>
> We could move it to mfd, but drivers in drivers/mfd which don't make use
> of MFD_CORE seems equally strange (although, I suppose there is
> precedent). Maybe Lee has some thoughts.

Is there any reason why the functionality can't be split up into
different source files? Create an LED driver, a Switch (whatever that
is) driver and a Watchdog driver, place them in drivers/{appropriate},
then register from each of them using the MFD API.

--
Lee Jones
Linaro STMicroelectronics Landing Team Lead
Linaro.org â Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog