Re: [PATCH v3 4/5] Makefile: Enable -Warray-bounds

[Kees Cook <keescook@chromium.org>]

On Mon, Aug 30, 2021 at 09:31:28PM -0700, Guenter Roeck wrote:
> On Fri, Aug 27, 2021 at 09:30:14AM -0700, Kees Cook wrote:
> > With the recent fixes for flexible arrays and expanded FORTIFY_SOURCE
> > coverage, it is now possible to enable -Warray-bounds. Since both
> > GCC and Clang include -Warray-bounds in -Wall, we just need to stop
> > disabling it.
> > 
> > Cc: Arnd Bergmann <arnd@arndb.de>
> > Cc: Masahiro Yamada <masahiroy@kernel.org>
> > Cc: linux-kbuild@vger.kernel.org
> > Co-developed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
> > Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
> > Signed-off-by: Kees Cook <keescook@chromium.org>
> > ---
> >  Makefile | 1 -
> >  1 file changed, 1 deletion(-)
> > 
> > diff --git a/Makefile b/Makefile
> > index e4f5895badb5..8e7e73a642e2 100644
> > --- a/Makefile
> > +++ b/Makefile
> > @@ -995,7 +995,6 @@ KBUILD_CFLAGS += $(call cc-disable-warning, stringop-truncation)
> >  
> >  # We'll want to enable this eventually, but it's not going away for 5.7 at least
> >  KBUILD_CFLAGS += $(call cc-disable-warning, zero-length-bounds)
> > -KBUILD_CFLAGS += $(call cc-disable-warning, array-bounds)
> >  KBUILD_CFLAGS += $(call cc-disable-warning, stringop-overflow)
> >  
> 
> This patch causes 'alpha' builds to fail when trying to build an image with
> gcc 11.2.
> 
> In file included from include/linux/string.h:20,
>                  from include/linux/bitmap.h:11,
>                  from include/linux/cpumask.h:12,
>                  from include/linux/smp.h:13,
>                  from include/linux/lockdep.h:14,
>                  from include/linux/spinlock.h:63,
>                  from include/linux/mmzone.h:8,
>                  from include/linux/gfp.h:6,
>                  from include/linux/mm.h:10,
>                  from include/linux/pagemap.h:8,
>                  from arch/alpha/mm/init.c:10:
> In function '__memset',
>     inlined from '__bad_pagetable' at arch/alpha/mm/init.c:79:2:
> arch/alpha/include/asm/string.h:37:32: error: '__builtin_memset' offset [0, 8191] is out of the bounds [0, 0] [-Werror=array-bounds]
>    37 |                         return __builtin_memset(s, c, n);
>       |                                ^~~~~~~~~~~~~~~~~~~~~~~~~
> In function '__memset',
>     inlined from '__bad_page' at arch/alpha/mm/init.c:86:2:
> arch/alpha/include/asm/string.h:37:32: error: '__builtin_memset' offset [0, 8191] is out of the bounds [0, 0] [-Werror=array-bounds]
>    37 |                         return __builtin_memset(s, c, n);
>       |                                ^~~~~~~~~~~~~~~~~~~~~~~~~
> In function '__memset',
>     inlined from 'paging_init' at arch/alpha/mm/init.c:256:2:
> arch/alpha/include/asm/string.h:37:32: error: '__builtin_memset' offset [0, 8191] is out of the bounds [0, 0] [-Werror=array-bounds]
>    37 |                         return __builtin_memset(s, c, n);
> 
> Seen in next-20210830.

Ah-ha, thanks for the report! I didn't see this in my builds -- what
config target did you use for this?

Looks like GCC isn't happy about an unsigned long->void * conversion
here...

include/asm/page.h:#define PAGE_OFFSET          0xfffffc0000000000UL
...
include/uapi/asm/setup.h:#define KERNEL_START_PHYS      0x1000000 /* required: Wildfire/Titan/Marvel */
include/uapi/asm/setup.h:#define EMPTY_PGT (PAGE_OFFSET+KERNEL_START_PHYS+0x04000)
...
mm/init.c:      memset((void *) EMPTY_PGT, 0, PAGE_SIZE);

I'll try to figure out the right annotations to fix this...

-- 
Kees Cook