build breakage from of i2c helper patch

build breakage from of i2c helper patch

[Kumar Gala]

I'm guessing the "[POWERPC] i2c: OF helpers for the i2c API"

http://git.kernel.org/?p=linux/kernel/git/paulus/powerpc.git;a=commitdiff;h=612212a3f2f053ea68ce9cd16d3deeca7754e8c9

is causing this build failure:

http://kisskb.ellerman.id.au/kisskb/buildresult/24375/

- k

Re: build breakage from of i2c helper patch

[Jochen Friedrich]

Hi Kumar,

> I'm guessing the "[POWERPC] i2c: OF helpers for the i2c API"
> 
> http://git.kernel.org/?p=linux/kernel/git/paulus/powerpc.git;a=commitdiff;h=612212a3f2f053ea68ce9cd16d3deeca7754e8c9
> 
> is causing this build failure:
> 
> http://kisskb.ellerman.id.au/kisskb/buildresult/24375/

It looks like it. i2c-core needs to be compiled static for the OF bindings to work.
I guess in Kconfig "depends on OF && I2C" must be changed into "depends on OF && (I2C=y)"

On the other hand, i didn't test the patch without the dependent patches, so it might not work as expected:

http://patchwork.ozlabs.org/linuxppc/patch?id=17833
http://patchwork.ozlabs.org/linuxppc/patch?id=17834

or the original ones from Jean Delvare:

http://patchwork.ozlabs.org/linuxppc/patch?id=16282
http://patchwork.ozlabs.org/linuxppc/patch?id=16283
http://patchwork.ozlabs.org/linuxppc/patch?id=16284

Thanks,
Jochen

Re: build breakage from of i2c helper patch

[Jean Delvare]

On Thu, 17 Apr 2008 11:25:12 +0200, Jochen Friedrich wrote:
> It looks like it. i2c-core needs to be compiled static for the OF bindings to work.
> I guess in Kconfig "depends on OF && I2C" must be changed into "depends on OF && (I2C=y)"

(blind shot) Maybe this means that some code that is in i2c-core should
be moved to i2c-boardinfo instead? i2c-boardinfo is always built into
the kernel even when i2c-core is modular.

-- 
Jean Delvare

Re: build breakage from of i2c helper patch

[Paul Mackerras]

Jochen Friedrich writes:

> It looks like it. i2c-core needs to be compiled static for the OF
> bindings to work.
> I guess in Kconfig "depends on OF && I2C" must be changed into
> "depends on OF && (I2C=y)"
> 
> On the other hand, i didn't test the patch without the dependent
> patches, so it might not work as expected:

Which way around is this dependency?  Do you mean that the patches you
listed depend on the "OF helpers for the i2c API" patch, or that the
OF helpers patch depends on these other ones?

> http://patchwork.ozlabs.org/linuxppc/patch?id=17833
> http://patchwork.ozlabs.org/linuxppc/patch?id=17834
> 
> or the original ones from Jean Delvare:
> 
> http://patchwork.ozlabs.org/linuxppc/patch?id=16282
> http://patchwork.ozlabs.org/linuxppc/patch?id=16283
> http://patchwork.ozlabs.org/linuxppc/patch?id=16284

What's the best way to fix this now?  We need to get a reasonable fix
in before I ask Linux to pull the powerpc tree.

Paul.

Re: build breakage from of i2c helper patch

[Jochen Friedrich]

Hi Jean,

> (blind shot) Maybe this means that some code that is in i2c-core should
> be moved to i2c-boardinfo instead? i2c-boardinfo is always built into
> the kernel even when i2c-core is modular.

unfortunately, i2c_new_device() tries to attach the module, so this function
really needs full i2c-core functionality.

I think the best way to fix this is to make the OF helpers def_tristate I2C, so
it will be compiled as module if i2c-core is compiled as module, as well.

I'll send an updated patch in a minute.

Thanks,
Jochen

Re: build breakage from of i2c helper patch

[Jochen Friedrich]

Hi Paul,

> Which way around is this dependency?  Do you mean that the patches you
> listed depend on the "OF helpers for the i2c API" patch, or that the
> OF helpers patch depends on these other ones?

The OF helpers patch depends on the patches below to support autoloading
i2c modules using the alias mechanism. Otherwise, we need to translate
between i2c module type and driver_name in the helper patch (like currently
done in fsl_soc.c function of_find_i2c_driver()). However, this old way
doesn't really scale well.

> 
>> http://patchwork.ozlabs.org/linuxppc/patch?id=17833
>> http://patchwork.ozlabs.org/linuxppc/patch?id=17834
>>
>> or the original ones from Jean Delvare:
>>
>> http://patchwork.ozlabs.org/linuxppc/patch?id=16282
>> http://patchwork.ozlabs.org/linuxppc/patch?id=16283
>> http://patchwork.ozlabs.org/linuxppc/patch?id=16284
> 
> What's the best way to fix this now?  We need to get a reasonable fix
> in before I ask Linux to pull the powerpc tree.

Thanks,
Jochen