[PATCH] dw_mmc: Add Synopsys DesignWare mmc host driver.

[Will Newton]

On Sun, Dec 12, 2010 at 2:11 PM, Russell King - ARM Linux
<linux at arm.linux.org.uk> wrote:
> On Sun, Dec 12, 2010 at 02:03:52PM +0000, Will Newton wrote:
>> On Sun, Dec 12, 2010 at 1:52 PM, Chris Ball <cjb at laptop.org> wrote:
>> > Hi Will,
>> >
>> > On Sun, Dec 12, 2010 at 10:57:44AM +0000, Will Newton wrote:
>> >> > drivers/mmc/host/dw_mmc.c: In function ‘dw_mci_pull_data64’:
>> >> > drivers/mmc/host/dw_mmc.c:998: error: implicit declaration of function ‘__raw_readq’
>> >> >
>> >> > because arch/arm doesn't implement raw versions of these 64-bit accesses.
>> >> > I'm surprised that this driver hasn't been compiled on ARM before!  What
>> >>
>> >> That particular bit of code has been added since it was last built for
>> >> arm. Our architecture can do 64bit accesses so we implement readq.
>> >> Unfortunately there doesn't seem to be a sane way to conditionalize
>> >> code for architectures that have or don't have readq, so I suspect
>> >> I'll have to just remove that branch of the if statement for now.
>> >
>> > (Russell, thanks for the excellent explanation.)
>> >
>> > Other drivers (MTD, gpio/basic_mmio_gpio.c, fs/fuse, pcm_oss.c)
>> > conditionalize uses of {read,write}q on BITS_PER_LONG >= 64, so
>> > something like this:
>>
>> I don't think that's going to work, BITS_PER_LONG isn't equivalent to
>> "can do 64bit IO accesses", at least it isn't on our architecture. x86
>> is in the same situation it would appear, although x86 does explicitly
>> #define readq so it may be possible to #ifdef on that?
>
> Maybe invent CONFIG_HAVE_MMIO_64BIT which architectures can select as
> appropriate?

Wouldn't it be simpler to have a fallback readq/writeq implementation
like the below?

static inline u64 __raw_readq(const volatile void __iomem *addr)
{
        return *(const volatile u64 __force *) addr;
}

It won't break any existing hardware (if your SoC bus does not support
64bit accesses you are unlikely to have peripherals that require it)
and would avoid a number of #ifdefs and/or Kconfig dependencies.

Currently asm-generic/io.h defines the above but it is guarded by
CONFIG_64BIT, which I am not sure is 100% correct in all situations
either.