INT19-C. Correctly compute integer widths

Integer types in C have two metrics associated with them, a size and a width. The size indicates the number of bytes used by a variable, and can be retrieved for any variable or type using the sizeof operator, while the width indicates the number of bits that are actually used to encode the integer's value. The C standard permits an integer to contain internal padding bits; these contribute to the integer's size, but not to its width. Consequently, an integer type may have a size that could be used to incorrectly assume its width is larger than it actually is.  Do not use the sizeof operator to compute the width of an integer type.

Noncompliant Code Example

This noncompliant code example illustrates a function that produces 2 raised to the power of the function argument. To prevent undefined behavior, in compliance with INT34-C. Do not shift a negative number of bits or more bits than exist in the operand, the function ensures that the argument is less than the number of bits used to store an unsigned int.

unsigned int pow2(unsigned int exp) {
  if (exp > sizeof( unsigned int) * CHAR_BIT) {
    /* handle error */
  }
  return 1 << exp;
}

Compliant Solution

This compliant solution eliminates the possibility of undefined behavior resulting from a left-shift operation on unsigned integers:

Modulo behavior resulting from left-shifting an unsigned integer type is permitted by this standard.

The computation of the correct width of any integer type can be nontrivial. On machines with no padding bits, the width can be computed directly from the integer's size:

Some platforms provide a population count operation, which counts the number of bits that are set. This can be used to compute the width:

For other platforms, you can implement _popcount yourself:

Noncompliant Code Example (Right Shift)

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 / 2E2. If E1 has a signed type and a negative value, the resulting value isimplementation-defined and can be either an arithmetic (signed) shift:

or a logical (unsigned) shift:

This noncompliant code example fails to test whether the right operand is greater than or equal to the width of the promoted left operand, allowing undefined behavior:

Making assumptions about whether a right shift is implemented as an arithmetic (signed) shift or a logical (unsigned) shift can also lead tovulnerabilities. See INT13-C. Use bitwise operators only on unsigned operands.

Compliant Solution (Right Shift)

This compliant solution tests the suspect shift operations to guarantee there is no possibility of undefined behavior:

Implementation Details

GCC has no options to handle shifts by negative amounts or by amounts outside the width of the type predictably or trap on them; they are always treated as undefined. Processors may reduce the shift amount modulo the width of the type. For example, 32-bit shifts are implemented using the following instructions on IA-32:

The sa[rl]l instructions take a bit mask of the least significant 5 bits from %cl to produce a value in the range [0, 31] and then shift %eax that many bits:

where %eax stores the least significant bits in the doubleword to be shifted, and %edx stores the most significant bits.

Risk Assessment

Although shifting a negative number of bits or more bits than exist in the operand is undefined behavior in C, the risk is generally low because processors frequently reduce the shift amount modulo the width of the type.

Automated Detection

Related Vulnerabilities

Search for vulnerabilities resulting from the violation of this rule on the CERT website.

Related Guidelines

Bibliography

A test program for this rule is available at www.securecoding.cert.org.