This rule finds code that checks the sign of the result of a bitwise operation. Such a check may yield unexpected results. As an example, consider the following code that checks if the nth bit of a variable x is set:

  x & (1 << n) > 0 

If x is a 32-bit signed integer, the value of x & (1 << 31) is interpreted as a signed number. If x is negative (that is, its sign bit is set), and n is 31, then x & (1 << 31) evaluates to 0x80000000 (all bits zero except the sign bit). The sign check on this value fails, implying that the 31st bit of x is unset. This is clearly incorrect.