Re: [PATCH 2/5] gpio/xilinx: enable for MIPS

["Sören Brinkmann" <soren.brinkmann@xilinx.com>]

On Wed, 2015-10-14 at 07:24PM +0200, Lars-Peter Clausen wrote:
> On 10/14/2015 06:54 PM, Sören Brinkmann wrote:
> > On Wed, 2015-10-14 at 05:57PM +0200, Lars-Peter Clausen wrote:
> >> On 10/14/2015 05:18 PM, Sören Brinkmann wrote:
> >>> On Wed, 2015-10-14 at 01:51PM +0100, Zubair Lutfullah Kakakhel wrote:
> >>>> MIPSfpga uses the axi gpio controller. Enable the driver for MIPS.
> >>>>
> >>>> Signed-off-by: Zubair Lutfullah Kakakhel <Zubair.Kakakhel@imgtec.com>
> >>>> ---
> >>>>  drivers/gpio/Kconfig | 2 +-
> >>>>  1 file changed, 1 insertion(+), 1 deletion(-)
> >>>>
> >>>> diff --git a/drivers/gpio/Kconfig b/drivers/gpio/Kconfig
> >>>> index 8949b3f..58e9afd 100644
> >>>> --- a/drivers/gpio/Kconfig
> >>>> +++ b/drivers/gpio/Kconfig
> >>>> @@ -508,7 +508,7 @@ config GPIO_XGENE_SB
> >>>>  
> >>>>  config GPIO_XILINX
> >>>>  	tristate "Xilinx GPIO support"
> >>>> -	depends on OF_GPIO && (PPC || MICROBLAZE || ARCH_ZYNQ || X86)
> >>>> +	depends on OF_GPIO && (PPC || MICROBLAZE || ARCH_ZYNQ || X86 || MIPS)
> >>>
> >>> Hmm, in general, this driver is hopefully generic enough that it doesn't
> >>> have any real architecture dependencies. And I suspect, we want to
> >>> enable this driver for ARM64 for ZynqMP soon too. Should we probably
> >>> drop these arch dependencies completely? It seems to become quite a long list.
> >>
> >> I've been thinking about this a while ago. This is certainly not the only
> >> driver affected by this problem. But the thing is people always complain if
> >> new symbols become visable in Kconfig that don't apply to their platform.
> >>
> >> Maybe we should introduce a HAS_REPROGRAMABLE_LOGIC (or similar) feature
> >> Kconfig symbol and let platforms which have a FPGA select it and let drivers
> >> for FPGA peripherals depend on it.
> > 
> > Sounds like a good idea to me. But, does that work for all use-cases.
> > E.g. if you plug some PCIe card with an FPGA into an x86(_64) machine.
> > That would allow you to use those drivers, but I'm not sure how that
> > could pull in the new config symbol.
> 
> Hm, right. We could also make it a user-selectable config symbol. In that
> case you only need to disable one symbol when you don't have FPGA support
> rather than one for each driver. Although I'm not quite sure where to put
> such a symbol.

Eventually, the FPGA manager subsystem could probably provide some high
level config symbols. Though, it is probably also not given that every
FPGA-enabled platform needs the FPGA manager.

	Sören