Re: [kbuild-all] [mmotm:master 19/222] arch/sh/kernel/cpu/sh2/clock-sh7619.o:undefined reference to `followparent_recalc'

[Randy Dunlap <rdunlap@infradead.org>]

On 4/3/19 5:20 PM, Rong Chen wrote:
> 
> On 4/3/19 10:46 PM, Randy Dunlap wrote:
>> On 4/3/19 12:09 AM, Rong Chen wrote:
>>> On 4/3/19 2:26 PM, Randy Dunlap wrote:
>>>> On 4/2/19 10:54 PM, kbuild test robot wrote:
>>>>> Hi Randy,
>>>>>
>>>>> It's probably a bug fix that unveils the link errors.
>>>>>
>>>>> tree:   git://git.cmpxchg.org/linux-mmotm.git master
>>>>> head:   03590d39c08e0f2969871a5efcf27a366c1e8c60
>>>>> commit: cffa367bb8abe4c1424e93e345c7d63844d1c5db [19/222] sh: fix multiple function definition build errors
>>>>> config: sh-allmodconfig (attached as .config)
>>>>> compiler: sh4-linux-gnu-gcc (Debian 7.2.0-11) 7.2.0
>>>>> reproduce:
>>>>>           wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
>>>>>           chmod +x ~/bin/make.cross
>>>>>           git checkout cffa367bb8abe4c1424e93e345c7d63844d1c5db
>>>>>           # save the attached .config to linux build tree
>>>>>           GCC_VERSION=7.2.0 make.cross ARCH=sh
>>>>>
>>>>> All errors (new ones prefixed by >>):
>>>>>
>>>>>>> arch/sh/kernel/cpu/sh2/clock-sh7619.o:(.data+0x1c): undefined reference to `followparent_recalc'
>>>>> ---
>>>>> 0-DAY kernel test infrastructure                Open Source Technology Center
>>>>> https://lists.01.org/pipermail/kbuild-all                   Intel Corporation
>>>>>
>>>> Hi,
>>>> I suspect that it's more of an invalid .config file.
>>>> How do you generate the .config files?  or is it a defconfig?
>>> the config file was generated by "make ARCH=sh allmodconfig"
>>>
>>>
>>>> Yes, I have seen this build error, but I was able to get around it
>>>> by modifying the .config file.  That's why I suspect that it may be
>>>> an invalid .config file.
>>> Can you share the fix steps? We'll take a look at it.
>> Hi,
>>
>> For this build error:
>>>> arch/sh/kernel/cpu/sh2/clock-sh7619.o:(.data+0x1c): undefined reference to `followparent_recalc'
>> the problem is with CONFIG_COMMON_CLK.  The COMMON_CLK framework does not
>> provide this API.  However, in arch/sh/boards/Kconfig, COMMON_CLK is always
>> selected by SH_DEVICE_TREE.  By disabling SH_DEVICE_TREE, the build
>> succeeds.
> 
> Thanks for the explanation, It seems SH_DEVICE_TREE was enabled by allmodconfig.
> does it mean it's a problem of allmodconfig? we thought kernel could be built successfully.

I think that there are multiple $arch + boards/platforms that are very
board-specific where allmodconfig doesn't make sense, but that's mostly an
opinion.  I haven't tried to prove it.


-- 
~Randy