On Mon, Apr 30, 2018 at 2:29 PM, Rasmus Villemoes
<linux@xxxxxxxxxxxxxxxxxx> wrote:
> On 2018-04-30 22:16, Matthew Wilcox wrote:
>> On Mon, Apr 30, 2018 at 12:02:14PM -0700, Kees Cook wrote:
>>>
>>> Getting the constant ordering right could be part of the macro
>>> definition, maybe? i.e.:
>>>
>>> static inline void *kmalloc_ab(size_t a, size_t b, gfp_t flags)
>>> {
>>> if (__builtin_constant_p(a) && a != 0 && \
>>> b > SIZE_MAX / a)
>>> return NULL;
>>> else if (__builtin_constant_p(b) && b != 0 && \
>>> a > SIZE_MAX / b)
>>> return NULL;
>>>
>>> return kmalloc(a * b, flags);
>>> }
>>
>> Ooh, if neither a nor b is constant, it just didn't do a check ;-( This
>> stuff is hard.
>>
>>> (I just wish C had a sensible way to catch overflow...)
>>
>> Every CPU I ever worked with had an "overflow" bit ... do we have a
>> friend on the C standards ctte who might figure out a way to let us
>> write code that checks it?
>
> gcc 5.1+ (I think) have the __builtin_OP_overflow checks that should
> generate reasonable code. Too bad there's no completely generic
> check_all_ops_in_this_expression(a+b*c+d/e, or_jump_here). Though it's
> hard to define what they should be checked against - probably would
> require all subexpressions (including the variables themselves) to have
> the same type.
>
> plug: https://lkml.org/lkml/2015/7/19/358
That's a very nice series. Why did it never get taken? It seems to do
the right things quite correctly.
Daniel, while this isn't a perfect solution, is this something you'd
use in graphics-land?
-Kees
--
Kees Cook
Pixel Security
