Re: drivers/watchdog/rave-sp-wdt.c:278: undefined reference to `nvmem_cell_get'

[Guenter Roeck <linux@roeck-us.net>]

On 02/13/2018 06:20 AM, Andrey Smirnov wrote:
> On Tue, Feb 13, 2018 at 1:30 AM, Lee Jones <lee.jones@linaro.org> wrote:
>> On Tue, 13 Feb 2018, kbuild test robot wrote:
>>
>>> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
>>> head:   178e834c47b0d01352c48730235aae69898fbc02
>>> commit: c3bb333457218ca4ed9553be47c0f567b4ef8a38 watchdog: Add RAVE SP watchdog driver
>>> date:   5 weeks ago
>>> config: x86_64-randconfig-u0-02131401 (attached as .config)
>>> compiler: gcc-5 (Debian 5.5.0-3) 5.4.1 20171010
>>> reproduce:
>>>          git checkout c3bb333457218ca4ed9553be47c0f567b4ef8a38
>>>          # save the attached .config to linux build tree
>>>          make ARCH=x86_64
>>>
>>> All errors (new ones prefixed by >>):
>>>
>>>     drivers/watchdog/rave-sp-wdt.o: In function `rave_sp_wdt_probe':
>>>>> drivers/watchdog/rave-sp-wdt.c:278: undefined reference to `nvmem_cell_get'
>>>>> drivers/watchdog/rave-sp-wdt.c:281: undefined reference to `nvmem_cell_read'
>>>>> drivers/watchdog/rave-sp-wdt.c:287: undefined reference to `nvmem_cell_put'
>>>     make[2]: *** [vmlinux] Error 1
>>
>> Looks like the RAVE Watchdog driver should select NVMEM?
>>
> 
> AFAICT, this is happening when RAVE SP core and watchdog drivers are
> built as built-in and NVMEM is selected as module. Looking at other
> users on NVMEM in the kernel is seems that the pattern to prevent this
> problem is to add "depends on NVMEM || !NVMEM" line to consumer's
> Kconfig entry.
> 
Makes sense, since the driver still works if NVMEM is not enabled at all.

> I'll submit a patch to that effect shortly.
> 
Thanks,
Guenter