drivers/mmc/host/sdhci-of-core.c on sparc64

[Grant Likely]

On Thu, Jan 13, 2011 at 6:12 PM, Chris Ball <cjb at laptop.org> wrote:
> Hi,
>
> On Thu, Jan 13, 2011 at 03:41:40PM -0800, Andrew Morton wrote:
>> > > drivers/mmc/host/sdhci-of-core.c:24:25: asm/machdep.h: No such file or directory
>> > > drivers/mmc/host/sdhci-of-core.c: In function `sdhci_of_wp_inverted':
>> > > drivers/mmc/host/sdhci-of-core.c:115: error: implicit declaration of function `machine_is'
>> > >
>> > > That code's been there for a while.  Did someone change Kconfig?
>> >
>> > Can you attach a .config?  <asm/machdep.h> is arch-specific, so I'd
>> > suggest that you're building on an unsupported arch.
>> >
>>
>> sparc64 allmodconfig.
>
> You're right, Andrew -- Rob (CC'd) changed the MMC Kconfig to build this
> driver on Sparc.  Mainline commit 236cdc7bc71 ("of: make drivers depend
> on CONFIG_OF instead of CONFIG_PPC_OF").
>
> Rob also posted a patch to devicetree-discuss@, on top of the one above
> ("mmc: sdhci-of: fix build on non-powerpc platforms"), to fix up the
> Sparc build by ifdef'ing for PPC inside the driver.  Grant merged the
> first patch but not the second, hence sparc64 allmodconfig is broken
> now.  The reason Grant didn't merge the second patch may be that Wolfram
> objected to #ifdef proliferation inside the driver.
>
> Options, as I see it:
>  * revert the commit such that MMC_SDHCI_OF once again depends on PPC_OF
>  * take the second patch as-is
>  * come up with a less-#ifdeffy second patch
>
> Wolfram, would appreciate your input on what we should do here.  Thanks,

I've applied the 2nd patch.  I had applied it to my test tree, and
indeed I replied saying I did, but I had a corrupt git tree event
shortly after applying it to my test branch and evidently lost track
of it.

It's been pushed out to my next-devicetree branch:

git://git.secretlab.ca/git/linux-2.6.git next-devicetree

I'll ask Linus to pull after doing some sanity testing.

g.