Question

Quoting the code for computing the integer absolute value (abs) without branching from http://graphics.stanford.edu/~seander/bithacks.html:

int v;           // we want to find the absolute value of v
unsigned int r;  // the result goes here 
int const mask = v >> sizeof(int) * CHAR_BIT - 1;

r = (v + mask) ^ mask;

Patented variation:

r = (v ^ mask) - mask;

What is CHAR_BIT and how use it?

Answer 1

You should be aware that this code depends on the implementation-defined behavior of right bitshift on signed types. gcc promises to always give the sane behavior (sign-bit-extension) but ISO C allows the implementation to zero-fill the upper bits.

One way around this problem:

#ifdef HAVE_SIGN_EXTENDING_BITSHIFT
int const mask = v >> sizeof(int) * CHAR_BIT - 1;
#else
int const mask = -((unsigned)v >> sizeof(int) * CHAR_BIT - 1);
#endif

Your Makefile or config.h etc. can define HAVE_SIGN_EXTENDING_BITSHIFT at build time depending on your platform.