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

[Kees Cook]

On Wed, Feb 02, 2022 at 08:09:03AM -0800, Guenter Roeck wrote:
> On Wed, Aug 18, 2021 at 01:11:17AM -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@xxxxxxxx>
> > Cc: Masahiro Yamada <masahiroy@xxxxxxxxxx>
> > Cc: linux-kbuild@xxxxxxxxxxxxxxx
> > Co-developed-by: Gustavo A. R. Silva <gustavoars@xxxxxxxxxx>
> > Signed-off-by: Gustavo A. R. Silva <gustavoars@xxxxxxxxxx>
> > Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx>
> > ---
> >  Makefile | 1 -
> >  1 file changed, 1 deletion(-)
> > 
> > diff --git a/Makefile b/Makefile
> > index a4aca01a4835..af22b83cede7 100644
> > --- a/Makefile
> > +++ b/Makefile
> > @@ -1072,7 +1072,6 @@ 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 += -Wno-array-bounds
> >  KBUILD_CFLAGS += $(call cc-disable-warning, stringop-overflow)
> >  
> >  # Another good warning that we'll want to enable eventually
> 
> alpha:defconfig:
> 
> 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);

Ugh, these all appear to be the same "gcc thinks a literal is an offset
from NULL" problem Arnd described:

https://lore.kernel.org/linux-hardening/CAK8P3a1CT+-3J7Ufv9TQwRiV4i_+FF=-+phwSjWafyGuPfG+AA@xxxxxxxxxxxxxx

What's annoying is that these values, if stored in a variable, aren't
treated that way:
https://pastebin.ubuntu.com/p/ZjZR2tzRpx/

I guess having a common way to convert from an integer (or literal) into
a void * that GCC doesn't freak out about would be generally nice.

-Kees

-- 
Kees Cook