Re: [m68k:master 1174/1174] arch/m68k/include/asm/string.h:72:25: warning: '__builtin_memcpy' forming offset 8 is out of the bounds [0, 7]

[Geert Uytterhoeven]

Hi Finn,

On Thu, Mar 7, 2019 at 10:42 PM Finn Thain <fthain@xxxxxxxxxxxxxxxxxxx> wrote:
> On Thu, 7 Mar 2019, Geert Uytterhoeven wrote:
> > On Thu, Mar 7, 2019 at 3:59 AM Finn Thain <fthain@xxxxxxxxxxxxxxxxxxx> wrote:
> > > On Tue, 5 Mar 2019, Andreas Schwab wrote:
> > > > On Mar 05 2019, Finn Thain <fthain@xxxxxxxxxxxxxxxxxxx> wrote:
> > > >
> > > > > interesting that the kernel's strlen implementation in
> > > > > include/linux/string.h can't achieve this.
> > > >
> > > > This implementation is only available if ARCH_HAS_FORTIFY_SOURCE.
> > > >
> > >
> > > I see. Perhaps we could add another definition to that file:
> > >
> > > #if !defined(__NO_FORTIFY) && defined(__OPTIMIZE__) && defined(CONFIG_FORTIFY_SOURCE)
> > > ...
> > > #else
> > > __FORTIFY_INLINE __kernel_size_t strlen(const char *p)
> > > {
> > >         return __builtin_strlen(p);
> > > }
> > > #endif
> > >
> > > I didn't test that.
>
> I've tested it now, it works too. This may be a better solution than
> defining a strlen macro.
>
> diff --git a/include/linux/string.h b/include/linux/string.h
> index 7927b875f80c..ec9c0a206bd3 100644
> --- a/include/linux/string.h
> +++ b/include/linux/string.h
> @@ -436,6 +436,13 @@ __FORTIFY_INLINE char *strcpy(char *p, const char *q)
>         return p;
>  }
>
> +#else
> +
> +__FORTIFY_INLINE __kernel_size_t strlen(const char *p)
> +{
> +       return __builtin_strlen(p);
> +}
> +
>  #endif
>
>  /**
>
> > But the following patch seems to work...
> > >
> > > diff --git a/arch/m68k/include/asm/string.h b/arch/m68k/include/asm/string.h
> > > index f759d944c449..3cff6b128ed3 100644
> > > --- a/arch/m68k/include/asm/string.h
> > > +++ b/arch/m68k/include/asm/string.h
> > > @@ -71,4 +71,6 @@ extern void *memset(void *, int, __kernel_size_t);
> > >  extern void *memcpy(void *, const void *, __kernel_size_t);
> > >  #define memcpy(d, s, n) __builtin_memcpy(d, s, n)
> > >
> > > +#define strlen(s) __builtin_strlen(s)
> >
> > Shouldn't you add
> >
> >     #define __HAVE_ARCH_STRLEN
> >
> > here...
> >
>
> No, the link fails because the compiler still emits some references to
> strlen().

Despite -ffreestanding?!?

> > > --- a/lib/string.c
> > > +++ b/lib/string.c
> > > @@ -472,6 +472,7 @@ char *strim(char *s)
> > >  EXPORT_SYMBOL(strim);
> > >
> > >  #ifndef __HAVE_ARCH_STRLEN
> > > +#undef strlen
> >
> > ... so you can drop this change?

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