Re: [Security-discuss] handling integer overflows

[Paul Eggert]

On 12/24/2012 08:31 AM, Nikos Mavrogiannopoulos wrote:
> Wouldn't it be nicer if these were inline functions that perform
> the multiplication and return an overflow flag? e.g.
> int safe_mul(uint *res, uint a, uint b)

That might be a useful thing to have, too.
But it would have a couple of problems that
INT_MULTIPLY_OVERFLOW does not have.  In C, one would need
a function like safe_mul for every integer type of interest:
int, unsigned, long, unsigned long, time_t, dev_t, size_t,
off_t, etc.  And when mixing types, one would have to worry about
overflow when converting A or B to uint, a problem
that would occur before safe_mul began executing and
therefore a problem that safe_mul could not diagnose.