[Cryptography] GCC bug 30475 (was Re: bounded pointers in C)

[Viktor Dukhovni]

On Thu, Apr 24, 2014 at 11:18:00PM -0400, Jerry Leichter wrote:

> I believe C99 guarantees that int's representation is either 1's
> or 2's complement.  That makes it easy to produce the maximum value
> for a signed type. Assume we know T is some signed integral type.

That may be a reasonable constraint on the target architecture in
my case, but:

> T max = ~(T)0;
> if (max == 0)
> {	// 1's complement
> 	T sign_bit = 1;
> 	while (sign_bit > 0)
> 		sign_bit <<= 1;

Left shift of signed quantities, is undefined and hostile compilers
are free to do as they please here.

-- 
	Viktor.