[PATCH 11/22] pci: lto: fix PREL32 relocations

[Sami Tolvanen]

On Wed, Jun 24, 2020 at 04:03:48PM -0700, Nick Desaulniers wrote:
> On Wed, Jun 24, 2020 at 3:50 PM kernel test robot <lkp at intel.com> wrote:
> >
> > Hi Sami,
> >
> > Thank you for the patch! Perhaps something to improve:
> >
> > [auto build test WARNING on 26e122e97a3d0390ebec389347f64f3730fdf48f]
> >
> > url:    https://github.com/0day-ci/linux/commits/Sami-Tolvanen/add-support-for-Clang-LTO/20200625-043816
> > base:    26e122e97a3d0390ebec389347f64f3730fdf48f
> > config: i386-alldefconfig (attached as .config)
> > compiler: gcc-9 (Debian 9.3.0-13) 9.3.0
> > reproduce (this is a W=1 build):
> >         # save the attached .config to linux build tree
> >         make W=1 ARCH=i386
> 
> Note: W=1 ^
> 
> >
> > If you fix the issue, kindly add following tag as appropriate
> > Reported-by: kernel test robot <lkp at intel.com>
> >
> > All warnings (new ones prefixed by >>):
> >
> >    In file included from arch/x86/kernel/pci-dma.c:9:
> > >> include/linux/compiler-gcc.h:72:45: warning: no previous prototype for '__UNIQUE_ID_via_no_dac190' [-Wmissing-prototypes]
> >       72 | #define __UNIQUE_ID(prefix) __PASTE(__PASTE(__UNIQUE_ID_, prefix), __COUNTER__)
> >          |                                             ^~~~~~~~~~~~
> >    include/linux/pci.h:1914:7: note: in definition of macro '___DECLARE_PCI_FIXUP_SECTION'
> >     1914 |  void stub(struct pci_dev *dev) { hook(dev); }   \
> >          |       ^~~~
> 
> Should `stub` be qualified as `static inline`? https://godbolt.org/z/cPBXxW
> Or should stub be declared in this header, but implemented in a .c
> file?  (I'm guessing the former, since the `hook` callback comes from
> the macro).

Does static inline guarantee that the compiler won't rename the symbol?
The purpose of this change is to have a stable symbol name, which we can
safely use in inline assembly.

Sami