Re: [PATCH] kbuild: link lib-y objects to vmlinux forcibly when CONFIG_MODULES=y

[Masahiro Yamada]

Hi Nick, David,

On Tue, Mar 10, 2020 at 4:59 AM Nick Desaulniers
<ndesaulniers@xxxxxxxxxx> wrote:
>
> On Mon, Mar 9, 2020 at 3:55 AM kbuild test robot <lkp@xxxxxxxxx> wrote:
> >
> > Hi Masahiro,
> >
> > I love your patch! Yet something to improve:
> >
> > [auto build test ERROR on kbuild/for-next]
> > [also build test ERROR on linux/master linus/master v5.6-rc5 next-20200306]
> > [if your patch is applied to the wrong git tree, please drop us a note to help
> > improve the system. BTW, we also suggest to use '--base' option to specify the
> > base tree in git format-patch, please see https://stackoverflow.com/a/37406982]
> >
> > url:    https://github.com/0day-ci/linux/commits/Masahiro-Yamada/kbuild-link-lib-y-objects-to-vmlinux-forcibly-when-CONFIG_MODULES-y/20200309-115312
> > base:   https://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild.git for-next
> > config: sparc-defconfig (attached as .config)
> > compiler: sparc-linux-gcc (GCC) 7.5.0
> > reproduce:
> >         wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
> >         chmod +x ~/bin/make.cross
> >         # save the attached .config to linux build tree
> >         GCC_VERSION=7.5.0 make.cross ARCH=sparc
> >
> > If you fix the issue, kindly add following tag
> > Reported-by: kbuild test robot <lkp@xxxxxxxxx>
> >
> > All errors (new ones prefixed by >>):
> >
> >    arch/sparc/lib/strlen.o: In function `strlen':
> > >> (.text+0x0): multiple definition of `strlen'
> >    lib/string.o:string.c:(.text+0x2b0): first defined here
>
> So this looks like a bug in 32b Sparc, that we should fix first.
>
> arch/sparc/lib/strlen.S provides a definition of strlen, but it does
> not define the preprocessor token __HAVE_ARCH_STRLEN to avoid multiple
> definitions from producing a link error.
>
> Or rather, __HAVE_ARCH_STRLEN is provided in include/asm/string_64.h,
> but not for 32b sparc.  arch/sparc/strlen.o is unconditionally
> required in lib-y in arch/sparc/lib/Makefile.
>
> Either:
> 1. arch/sparc/lib/strlen.S supports 32b sparc, then __HAVE_ARCH_STRLEN
> and a forward declaration of strlen should be provided in
> arch/sparc/include/asm/string.h rather than
> arch/sparc/include/asm/string_64.h, or...
> 2. arch/sparc/lib/strlen.S does not support 32b sparc, then the
> inclusion of strlen.o in arch/sparc/lib/Makefile should be predicated
> on CONFIG_SPARC64.
>
> + Dave who maybe can provide guidance on how to proceed?  The use of
> the BRANCH32 macro in arch/sparc/lib/strlen.S seems to have different
> definitions based on CONFIG_SPARC64 vs CONFIG_SPARC32, which makes me
> thing it's case 1 above, but I'm not familiar with Sparc assembly to
> be confident.



I agree.
The former is probably the intention
according to this commit:

commit ae984d72e0632782dd98c3fcf469b9045ad0d449
Author: David S. Miller <davem@xxxxxxxxxxxxx>
Date:   Tue Dec 9 01:07:09 2008 -0800

    sparc: Unify strlen assembler.

    Use the new asm/asm.h header to help commonize the
    strlen assembler between 32-bit and 64-bit

    While we're here, use proper linux/linkage.h macros
    instead of by-hand stuff.

    Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx>



-- 
Best Regards
Masahiro Yamada