shifting in bitmasks

[manuel-simon]

Hi,

I have some problems when bitmasks are used to get data out of packets.

The problem is that shifting a signed/negative number is compiler-dependent. Some compilers perform an arithmetic shift if the value is signed/negative. ~0 can be interpreted as negative since the first bit is set. Then the shift is performed, filling up with 1 bits.

Standard C99 §6.5.7 (e.g. here):

The result of E1 >> E2 is E1 right-shifted E2 bit positions. If E1 has an unsigned type
or if E1 has a signed type and a nonnegative value, the value of the result is the integral
part  of  the  quotient  of E1 / 2 E2. If E1 has  a  signed  type  and  a  negative  value,  the
resulting value is implementation-defined.

What we need in the masks are logical shifts. These are performed if the value is positive/unsigned. This should be ensured declaring the 0 as an unsigned literal.
~0 -> ~0U