Search code examples
csizelimitstandards

C - Going over char limit - standard rules

This code:

signed char a = 128;
a++;
printf("%d", a);

Prints value of '-127'. I understand why that is, basically the value "resets" when it reaches the limit and goes from there, but I have trouble finding out if that is specified in standard, or it's just a random action my compiler does?

Solution

  • Assuming the most common platforms with 8 bit char types and 2's complement (x86, ARM, MIPS, PPC, MSP430, etc.), the following happens:

    • 128 is too large for a signed char. It is converted in an implementation defined way to a signed char. Typically it is simply truncated and bit-copied (1:1). In 2's complement, 0b1000000 is the 2's complement representation of decimal -128 in a signed char.
    • The following addition is basic maths: -128 + 1 -> -127

    On other platforms, the result varies, but most likely they just use the lower bits of the int constant 128.

    To detect such flaws (but not all), enable compiler warnings and pay heed to them. gcc will report a truncation warning (-Wconversion) for the initialiser.