unsigned int x = (unsigned int)a[0] << 24 | (unsigned int)a[1] << 16 | (unsigned int)a[2] << 8 | (unsigned int)a[3];
x = x << 30 | x >> 2;

互換性と簡潔さのバランスで、こんな感じじゃない？

unsigned intが32bit
charが8bit
限定