Re: [PATCH v1 net-next 4/7] ptp_pch: Switch to use module_pci_driver() macro

On Fri, Aug 13, 2021 at 9:23 PM Jakub Kicinski [off-list ref] wrote:

On Fri, 13 Aug 2021 18:39:14 +0300 Andy Shevchenko wrote:

quoted

On Fri, Aug 13, 2021 at 06:13:21PM +0300, Andy Shevchenko wrote:

quoted

On Fri, Aug 13, 2021 at 10:34:17PM +0800, kernel test robot wrote:

quoted

Hi Andy,

I love your patch! Yet something to improve:

[auto build test ERROR on net-next/master]

url:    https://github.com/0day-ci/linux/commits/Andy-Shevchenko/ptp_pch-use-mac_pton/20210813-203135
base:   https://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next.git b769cf44ed55f4b277b89cf53df6092f0c9082d0
config: nios2-randconfig-r023-20210813 (attached as .config)
compiler: nios2-linux-gcc (GCC) 11.2.0
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/0day-ci/linux/commit/6c1fff5c80fe8f1a12c20bac2d28ebfa5960bde7
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Andy-Shevchenko/ptp_pch-use-mac_pton/20210813-203135
        git checkout 6c1fff5c80fe8f1a12c20bac2d28ebfa5960bde7
        # save the attached .config to linux build tree
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-11.2.0 make.cross ARCH=nios2

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <redacted>

Thanks!

Definitely I have compiled it in my local branch. I'll check what is the root
cause of this.

Kconfig misses PCI dependency. I will send a separate patch, there is nothing
to do here.

That patch has to be before this one, tho.

Yes, I have sent it as a fix to the net, this series is to the
net-next. Am I missing something in the process here? Because I have
been told a few times that I mustn't collect net and net-next patches
in one series in the usual cases.

There is a static inline
stub for pci_register_driver() etc. if !PCI, but there isn't for
module_pci_driver(), meaning in builds without PCI this driver used
to be harmlessly pointless, now it's breaking build.

Am I missing something?

Adding Bjorn in case he has a preference on adding the dependency vs
stubbing out module_pci_driver().

I went all the same way but I have checked SPI and I²C how they are
doing. It seems the common practice is to use direct dependency.
Nevertheless, the stubs in PCI puzzled me a bit, why do we have them
in the first place?


-- 
With Best Regards,
Andy Shevchenko