Re: [PATCH v11.5] overflow: Introduce overflows_type() and __castable_to_type()

[Kees Cook <keescook@chromium.org>]

On Mon, Sep 26, 2022 at 06:57:53PM +0300, Gwan-gyeong Mun wrote:
> 
> 
> On 9/26/22 3:37 AM, Kees Cook wrote:
> > Add overflows_type() to test if a variable or constant value would
> > overflow another variable or type. This can be used as a constant
> > expression for static_assert() (which requires a constant
> > expression[1][2]) when used on constant values. This must be constructed
> > manually, since __builtin_add_overflow() does not produce a constant
> > expression[3].
> > 
> > Additionally adds __castable_to_type(), similar to __same_type(), for
> > checking if a constant value will fit in a given type (i.e. it could
> > be cast to the type without overflow).
> > 
> > Add unit tests for overflows_type(), __same_type(), and
> > __castable_to_type() to the existing KUnit "overflow" test.
> > 
> > [1] https://en.cppreference.com/w/c/language/_Static_assert
> > [2] C11 standard (ISO/IEC 9899:2011): 6.7.10 Static assertions
> > [3] https://gcc.gnu.org/onlinedocs/gcc/Integer-Overflow-Builtins.html
> >      6.56 Built-in Functions to Perform Arithmetic with Overflow Checking
> >      Built-in Function: bool __builtin_add_overflow (type1 a, type2 b,
> >                                                      type3 *res)
> > 
> > Cc: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
> > Cc: Nathan Chancellor <nathan@kernel.org>
> > Cc: Nick Desaulniers <ndesaulniers@google.com>
> > Cc: Tom Rix <trix@redhat.com>
> > Cc: Daniel Latypov <dlatypov@google.com>
> > Cc: Vitor Massaru Iha <vitor@massaru.org>
> > Cc: "Gustavo A. R. Silva" <gustavoars@kernel.org>
> > Cc: linux-hardening@vger.kernel.org
> > Cc: llvm@lists.linux.dev
> > Co-developed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
> > Signed-off-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
> > Signed-off-by: Kees Cook <keescook@chromium.org>
> > ---
> >   include/linux/compiler.h |   1 +
> >   include/linux/overflow.h |  48 +++++
> >   lib/overflow_kunit.c     | 393 ++++++++++++++++++++++++++++++++++++++-
> >   3 files changed, 441 insertions(+), 1 deletion(-)
> > 
> > diff --git a/include/linux/compiler.h b/include/linux/compiler.h
> > index 7713d7bcdaea..c631107e93b1 100644
> > --- a/include/linux/compiler.h
> > +++ b/include/linux/compiler.h
> > @@ -244,6 +244,7 @@ static inline void *offset_to_ptr(const int *off)
> >    * bool and also pointer types.
> >    */
> >   #define is_signed_type(type) (((type)(-1)) < (__force type)1)
> > +#define is_unsigned_type(type) (!is_signed_type(type))
> >   /*
> >    * This is needed in functions which generate the stack canary, see
> > diff --git a/include/linux/overflow.h b/include/linux/overflow.h
> > index 19dfdd74835e..c8cbeae5f4f8 100644
> > --- a/include/linux/overflow.h
> > +++ b/include/linux/overflow.h
> > @@ -127,6 +127,54 @@ static inline bool __must_check __must_check_overflow(bool overflow)
> >   	(*_d >> _to_shift) != _a);					\
> >   }))
> > +#define __overflows_type_constexpr(x, T) (			\
> > +	is_unsigned_type(typeof(x)) ?				\
> > +		(x) > type_max(typeof(T)) ? 1 : 0		\
> > +	: is_unsigned_type(typeof(T)) ?				\
> > +		(x) < 0 || (x) > type_max(typeof(T)) ? 1 : 0	\
> > +		: (x) < type_min(typeof(T)) ||			\
> > +		  (x) > type_max(typeof(T)) ? 1 : 0 )
> > +
> > +#define __overflows_type(x, T)		({	\
> > +	typeof(T) v = 0;			\
> > +	check_add_overflow((x), v, &v);		\
> > +})
> > +
> > +/**
> > + * overflows_type - helper for checking the overflows between value, variables,
> > + *		    or data type
> > + *
> > + * @n: source constant value or variable to be checked
> > + * @T: destination variable or data type proposed to store @x
> > + *
> > + * Compares the @x expression for whether or not it can safely fit in
> > + * the storage of the type in @T. @x and @T can have different types.
> > + * If @x is a conxtant expression, this will also resolve to a constant
> > + * expression.
> > + *
> > + * Returns: true if overflow can occur, false otherwise.
> > + */
> > +#define overflows_type(n, T)					\
> > +	__builtin_choose_expr(__is_constexpr(n),		\
> > +			      __overflows_type_constexpr(n, T),	\
> > +			      __overflows_type(n, T))
> > +
> > +/**
> > + * __castable_to_type - like __same_type(), but also allows for casted literals
> > + *
> > + * @n: variable or constant value
> > + * @T: data type or variable
> > + *
> > + * Unlike the __same_type() macro, this allows a constant value as the
> > + * first argument. If this value would not overflow into an assignment
> > + * of the second argument's type, it returns true. Otherwise, this falls
> > + * back to __same_type().
> > + */
> > +#define __castable_to_type(n, T)					\
> > +	__builtin_choose_expr(__is_constexpr(n),			\
> > +			      !__overflows_type_constexpr(n, T),	\
> > +			      __same_type(n, T))
> > +
> This name is fine, but I prefer the __same_typable you suggested as a
> comment in the previous patch better, what do you think?
> ( __castable_to_type(n, T); The macro name seems to handle if type casting
> is possible to the second argument type from the first argument variable. )

I changed this name because "typable" isn't a familiar name for someone
reading all of this for the first time. What's really happening is a
check if _casting_ will result in an overflow. And when I named it just
"__castable_type" it sounded like a declaration rather than a test. But
perhaps it should lose the "__" prefix, and just be "castable_to_type"?
Or even more verbose as "can_cast_to_type()" ?

As for argument order, it seemed best to keep the order the same as with
overflows_type(). I think that makes all of these macros a bit easier to
read/review/understand for others.

-- 
Kees Cook