Re: [PATCH] kbuild: fix kernel/bounds.c 'W=1' warning

From: Geert Uytterhoeven
Date: Mon Oct 08 2018 - 10:41:17 EST


On Mon, Oct 8, 2018 at 4:33 PM Masahiro Yamada
<yamada.masahiro@xxxxxxxxxxxxx> wrote:
> On Mon, Oct 8, 2018 at 7:01 PM Kieran Bingham
> <kieran.bingham+renesas@xxxxxxxxxxxxxxxx> wrote:
> > On 06/10/18 23:06, Masahiro Yamada wrote:
> > > On Sun, Oct 7, 2018 at 6:58 AM Masahiro Yamada
> > > <yamada.masahiro@xxxxxxxxxxxxx> wrote:
> > >>
> > >> Hi Miguel,
> > >>
> > >>
> > >> On Sun, Oct 7, 2018 at 6:18 AM Miguel Ojeda
> > >> <miguel.ojeda.sandonis@xxxxxxxxx> wrote:
> > >>>
> > >>> On Fri, Oct 5, 2018 at 10:35 AM Arnd Bergmann <arnd@xxxxxxxx> wrote:
> > >>>>
> > >>>> Building any configuration with 'make W=1' produces a warning:
> > >>>>
> > >>>> kernel/bounds.c:16:6: warnign: no previous prototype for 'foo' [-Wmissing-prototypes]
> > >>>>
> > >>>> When also passing -Werror, this prevents us from building any
> > >>>> other files. Nobody ever calls the function, but we can't make
> > >>>> it 'static' either since we want the compiler output.
> > >>>>
> > >>>> Calling it 'main' instead however avoids the warning, because gcc
> > >>>> does not insist on having a declaration for main.
> > >>>
> > >>> I think marking the function as static __used should do the trick and
> > >>> would be less confusing.
> > >>
> > >>
> > >>
> > >>
> > >> I tried __used, but I still see the warning.
> > >>
> > >>
> > >> masahiro@grover:~/ref/linux$ git diff
> > >> diff --git a/kernel/bounds.c b/kernel/bounds.c
> > >> index c373e88..aee0101 100644
> > >> --- a/kernel/bounds.c
> > >> +++ b/kernel/bounds.c
> > >> @@ -13,7 +13,7 @@
> > >> #include <linux/log2.h>
> > >> #include <linux/spinlock_types.h>
> > >>
> > >> -void foo(void)
> > >> +void __used foo(void)
> > >> {
> > >> /* The enum constants to put into include/generated/bounds.h */
> > >> DEFINE(NR_PAGEFLAGS, __NR_PAGEFLAGS);
> > >> masahiro@grover:~/ref/linux$ make W=1 prepare
> > >> CC kernel/bounds.s
> > >> kernel/bounds.c:16:13: warning: no previous prototype for âfooâ
> > >> [-Wmissing-prototypes]
> > >> void __used foo(void)
> > >> ^
> > >> CC arch/x86/kernel/asm-offsets.s
> > >
> > >
> > >
> > > Sorry, I forgot to add 'static'.
> > >
> > > Adding both static and __used worked for me,
> > > and I like the idea.
> > >
> >
> > Aha - I'd also tried converting to static in my earlier attempts, but
> > didn't realise we had __used!
> >
> > updating as "static __used" causes the following diff:
> >
> > diff -Nurp bounds.s.foo bounds.s.static-used
> > --- bounds.s.foo 2018-10-05 10:20:53.269941404 +0100
> > +++ bounds.s.static-used 2018-10-08 10:51:18.079309049 +0100
> > @@ -110,7 +110,6 @@
> > #NO_APP
> > .align 2
> > .p2align 3,,7
> > - .global foo
> > .type foo, %function
> > foo:
> > stp x29, x30, [sp, -16]! //,,,
> >
> >
> > I'd say this is a pretty good alternative fix - however I see Arnd's
> > version is already on it's way though akpm's tree...
> >
> > https://ozlabs.org/~akpm/mmots/broken-out/kbuild-fix-kernel-boundsc-w%3D1-warning.patch
> >
> > Anyway, as long as one of the variants gets there I'll be happy :)
>
>
> I will leave it to Arnd.
>
>
> When we fix arch/{mips,sparc}/kernel/asm-offsets.c,
> 'static __used' is just additions.
>
> The 'main(void)' solution would require a little bit restructuring.
>
>
>
> FWIW, with my quick analysis, the following should be fixed as well:
>
> arch/alpha/kernel/asm-offsets.c
> arch/c6x/kernel/asm-offsets.c
> arch/ia64/kernel/asm-offsets.c
> arch/ia64/kernel/nr-irqs.c
> arch/mips/kernel/asm-offsets.c
> arch/riscv/kernel/asm-offsets.c
> arch/sparc/kernel/asm-offsets.c
> arch/x86/kernel/asm-offsets.c
> arch/x86/um/shared/sysdep/kernel-offsets.h
> samples/bpf/syscall_nrs.c

All the other asm-offsets.c files already use "int main(void)" ;-)

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds