Re: mmotm 2019-08-27-20-39 uploaded (objtool: xen)

From: Josh Poimboeuf <jpoimboe@redhat.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Randy Dunlap <rdunlap@infradead.org>,
	akpm@linux-foundation.org, broonie@kernel.org,
	linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org, linux-next@vger.kernel.org, mhocko@suse.cz,
	mm-commits@vger.kernel.org, sfr@canb.auug.org.au
Subject: Re: mmotm 2019-08-27-20-39 uploaded (objtool: xen)
Date: Thu, 29 Aug 2019 18:37:35 -0500	[thread overview]
Message-ID: <20190829233735.yp3mwhg6er353qw5@treble> (raw)
In-Reply-To: <20190829082445.GM2369@hirez.programming.kicks-ass.net>

On Thu, Aug 29, 2019 at 10:24:45AM +0200, Peter Zijlstra wrote:
> On Wed, Aug 28, 2019 at 03:01:34PM -0500, Josh Poimboeuf wrote:
> > On Wed, Aug 28, 2019 at 10:56:25AM -0700, Randy Dunlap wrote:
> > > >> drivers/xen/gntdev.o: warning: objtool: gntdev_copy()+0x229: call to __ubsan_handle_out_of_bounds() with UACCESS enabled
> > > > 
> > > > Easy one :-)
> > > > 
> > > > diff --git a/tools/objtool/check.c b/tools/objtool/check.c
> > > > index 0c8e17f946cd..6a935ab93149 100644
> > > > --- a/tools/objtool/check.c
> > > > +++ b/tools/objtool/check.c
> > > > @@ -483,6 +483,7 @@ static const char *uaccess_safe_builtin[] = {
> > > >  	"ubsan_type_mismatch_common",
> > > >  	"__ubsan_handle_type_mismatch",
> > > >  	"__ubsan_handle_type_mismatch_v1",
> > > > +	"__ubsan_handle_out_of_bounds",
> > > >  	/* misc */
> > > >  	"csum_partial_copy_generic",
> > > >  	"__memcpy_mcsafe",
> > > > 
> > > 
> > > 
> > > then I get this one:
> > > 
> > > lib/ubsan.o: warning: objtool: __ubsan_handle_out_of_bounds()+0x5d: call to ubsan_prologue() with UACCESS enabled
> > 
> > And of course I jinxed it by calling it easy.
> > 
> > Peter, how do you want to handle this?
> > 
> > Should we just disable UACCESS checking in lib/ubsan.c?
> 
> No, that is actually unsafe and could break things (as would you patch
> above).

Oops.  -EFIXINGTOOMANYOBJTOOLISSUESATONCE

> I'm thinking the below patch ought to cure things:
> 
> ---
> Subject: x86/uaccess: Don't leak the AC flags into __get_user() argument evalidation

s/evalidation/evaluation

> Identical to __put_user(); the __get_user() argument evalution will too
> leak UBSAN crud into the __uaccess_begin() / __uaccess_end() region.
> While uncommon this was observed to happen for:
> 
>   drivers/xen/gntdev.c: if (__get_user(old_status, batch->status[i]))
> 
> where UBSAN added array bound checking.
> 
> This complements commit:
> 
>   6ae865615fc4 ("x86/uaccess: Dont leak the AC flag into __put_user() argument evaluation")
> 
> Reported-by: Randy Dunlap <rdunlap@infradead.org>
> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> Cc: luto@kernel.org
> ---
>  arch/x86/include/asm/uaccess.h | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/x86/include/asm/uaccess.h b/arch/x86/include/asm/uaccess.h
> index 9c4435307ff8..35c225ede0e4 100644
> --- a/arch/x86/include/asm/uaccess.h
> +++ b/arch/x86/include/asm/uaccess.h
> @@ -444,8 +444,10 @@ __pu_label:							\
>  ({									\
>  	int __gu_err;							\
>  	__inttype(*(ptr)) __gu_val;					\
> +	__typeof__(ptr) __gu_ptr = (ptr);				\
> +	__typeof__(size) __gu_size = (size);				\
>  	__uaccess_begin_nospec();					\
> -	__get_user_size(__gu_val, (ptr), (size), __gu_err, -EFAULT);	\
> +	__get_user_size(__gu_val, __gu_ptr, __gu_size, __gu_err, -EFAULT);	\
>  	__uaccess_end();						\
>  	(x) = (__force __typeof__(*(ptr)))__gu_val;			\
>  	__builtin_expect(__gu_err, 0);					\

Reviewed-by: Josh Poimboeuf <jpoimboe@redhat.com>

-- 
Josh