broken dependency of drivers/mtd/maps/gpio-addr-flash.o

[Mike Frysinger]

On Sat, Oct 10, 2009 at 13:18, David Woodhouse wrote:
> On Sat, 10 Oct 2009, Mike Frysinger wrote:
>> On Sat, Oct 10, 2009 at 12:19, David Woodhouse wrote:
>>> On Sat, 10 Oct 2009, Mike Frysinger wrote:
>>>> On Sat, Oct 10, 2009 at 10:56, Michal Suchanek wrote:
>>>>> It tried building the 2.6.32-rc3 kernel for i386.
>>>>>
>>>>> This file fails to compile if gpiolib is not configured but it is
>>>>> possible to select compilation of that file without selecting gpiolib.
>>>>>
>>>>> I guess I do not need this file because I do not have gpio and it
>>>>> really depends on it but drivers in the MTD section seemed to be
>>>>> mostly enabled in distribution config and this was a new driver since
>>>>> 2.6.30.
>>>>
>>>> patches have already been posted for the issue
>>>
>>> Have they? I thought the x86 gpio implementation was still broken?
>>
>> i'm talking about the gpio driver adding depend lines for gpio
>
> Not in front of a real computer right now but I think you added a dep on
> GENERIC_GPIO || GPIOLIB, but it fails on x86 when the former is enabled
> (which it always has) but not the latter (which is optional).
> x86 needs _both_ defined.

yes, the issue you raised in the original thread has not been
addressed, but until the gpio-addr-flash Kconfig itself is fixed, it's
hard to filter out problems people are having between x86-is-broken
and gpio-addr-flash-is-broken.  i'm addressing the latter and then we
can move on to complaining to the x86 people to fix their side.
-mike