[PATCH v7 1/8] overflow: Move and add few utility macros into overflow

[Kees Cook]

On Thu, Aug 18, 2022 at 01:07:29AM +0200, Andi Shyti wrote:
> Hi Kees,
> 
> would you mind taking a look at this patch?

Hi! Thanks for the heads-up!

> 
> Thanks,
> Andi
> 
> On Tue, Aug 16, 2022 at 06:35:18PM +0900, Gwan-gyeong Mun wrote:
> > It moves overflows_type utility macro into overflow header from i915_utils
> > header. The overflows_type can be used to catch the truncation between data
> > types. And it adds safe_conversion() macro which performs a type conversion
> > (cast) of an source value into a new variable, checking that the
> > destination is large enough to hold the source value. And the functionality
> > of overflows_type has been improved to handle the signbit.
> > The is_unsigned_type macro has been added to check the sign bit of the
> > built-in type.
> > 
> > v3: Add is_type_unsigned() macro (Mauro)
> >     Modify overflows_type() macro to consider signed data types (Mauro)
> >     Fix the problem that safe_conversion() macro always returns true
> > v4: Fix kernel-doc markups
> > v6: Move macro addition location so that it can be used by other than drm
> >     subsystem (Jani, Mauro, Andi)
> >     Change is_type_unsigned to is_unsigned_type to have the same name form
> >     as is_signed_type macro
> > 
> > Signed-off-by: Gwan-gyeong Mun <gwan-gyeong.mun at intel.com>
> > Cc: Thomas Hellström <thomas.hellstrom at linux.intel.com>
> > Cc: Matthew Auld <matthew.auld at intel.com>
> > Cc: Nirmoy Das <nirmoy.das at intel.com>
> > Cc: Jani Nikula <jani.nikula at intel.com>
> > Cc: Andi Shyti <andi.shyti at linux.intel.com>
> > Reviewed-by: Mauro Carvalho Chehab <mchehab at kernel.org> (v5)
> > ---
> >  drivers/gpu/drm/i915/i915_utils.h |  5 +--
> >  include/linux/overflow.h          | 54 +++++++++++++++++++++++++++++++
> >  2 files changed, 55 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/i915_utils.h b/drivers/gpu/drm/i915/i915_utils.h
> > index c10d68cdc3ca..eb0ded23fa9c 100644
> > --- a/drivers/gpu/drm/i915/i915_utils.h
> > +++ b/drivers/gpu/drm/i915/i915_utils.h
> > @@ -32,6 +32,7 @@
> >  #include <linux/types.h>
> >  #include <linux/workqueue.h>
> >  #include <linux/sched/clock.h>
> > +#include <linux/overflow.h>
> >  
> >  #ifdef CONFIG_X86
> >  #include <asm/hypervisor.h>
> > @@ -111,10 +112,6 @@ bool i915_error_injected(void);
> >  #define range_overflows_end_t(type, start, size, max) \
> >  	range_overflows_end((type)(start), (type)(size), (type)(max))
> >  
> > -/* Note we don't consider signbits :| */
> > -#define overflows_type(x, T) \
> > -	(sizeof(x) > sizeof(T) && (x) >> BITS_PER_TYPE(T))
> > -
> >  #define ptr_mask_bits(ptr, n) ({					\
> >  	unsigned long __v = (unsigned long)(ptr);			\
> >  	(typeof(ptr))(__v & -BIT(n));					\
> > diff --git a/include/linux/overflow.h b/include/linux/overflow.h
> > index f1221d11f8e5..462a03454377 100644
> > --- a/include/linux/overflow.h
> > +++ b/include/linux/overflow.h
> > @@ -35,6 +35,60 @@
> >  #define type_max(T) ((T)((__type_half_max(T) - 1) + __type_half_max(T)))
> >  #define type_min(T) ((T)((T)-type_max(T)-(T)1))
> >  
> > +/**
> > + * is_unsigned_type - helper for checking data type which is an unsigned data
> > + * type or not
> > + * @x: The data type to check
> > + *
> > + * Returns:
> > + * True if the data type is an unsigned data type, false otherwise.
> > + */
> > +#define is_unsigned_type(x) ((typeof(x))-1 >= (typeof(x))0)

I'd rather not have separate logic for this. Instead, I'd like it to be:

#define is_unsigned_type(x) (!is_signed_type(x))

> > +
> > +/**
> > + * overflows_type - helper for checking the truncation between data types
> > + * @x: Source for overflow type comparison
> > + * @T: Destination for overflow type comparison
> > + *
> > + * It compares the values and size of each data type between the first and
> > + * second argument to check whether truncation can occur when assigning the
> > + * first argument to the variable of the second argument.
> > + * Source and Destination can be used with or without sign bit.
> > + * Composite data structures such as union and structure are not considered.
> > + * Enum data types are not considered.
> > + * Floating point data types are not considered.
> > + *
> > + * Returns:
> > + * True if truncation can occur, false otherwise.
> > + */
> > +#define overflows_type(x, T) \
> > +	(is_unsigned_type(x) ? \
> > +		is_unsigned_type(T) ? \
> > +			(sizeof(x) > sizeof(T) && (x) >> BITS_PER_TYPE(T)) ? 1 : 0 \
> > +			: (sizeof(x) >= sizeof(T) && (x) >> (BITS_PER_TYPE(T) - 1)) ? 1 : 0 \
> > +	: is_unsigned_type(T) ? \
> > +		((x) < 0) ? 1 : (sizeof(x) > sizeof(T) && (x) >> BITS_PER_TYPE(T)) ? 1 : 0 \
> > +		: (sizeof(x) > sizeof(T)) ? \
> > +			((x) < 0) ? (((x) * -1) >> BITS_PER_TYPE(T)) ? 1 : 0 \
> > +				: ((x) >> BITS_PER_TYPE(T)) ? 1 : 0 \
> > +			: 0)

Like the other, I'd much rather this was rephrased in terms of the
existing macros (e.g. type_min()/type_max().)

> > +
> > +/**
> > + * safe_conversion - perform a type conversion (cast) of an source value into
> > + * a new variable, checking that the destination is large enough to hold the
> > + * source value.
> > + * @ptr: Destination pointer address
> > + * @value: Source value
> > + *
> > + * Returns:
> > + * If the value would overflow the destination, it returns false.
> > + */
> > +#define safe_conversion(ptr, value) ({ \
> > +	typeof(value) __v = (value); \
> > +	typeof(ptr) __ptr = (ptr); \
> > +	overflows_type(__v, *__ptr) ? 0 : ((*__ptr = (typeof(*__ptr))__v), 1); \
> > +})

I try to avoid "safe" as an adjective for interface names, since it
doesn't really answer "safe from what?" This looks more like "assign, but
zero when out of bounds". And it can be built from existing macros here:

	if (check_add_overflow(0, value, ptr))
		*ptr = 0;

I actually want to push back on this a bit, because there can still be
logic bugs built around this kind of primitive. Shouldn't out-of-bounds
assignments be seen as a direct failure? I would think this would be
sufficient:

#define check_assign(value, ptr)	check_add_overflow(0, value, ptr)

And callers would do:

	if (check_assign(value, &var))
		return -EINVAL;

etc.


-- 
Kees Cook