INT05-C. Do not use input functions to convert character data if they cannot handle all possible inputs

Do not use functions that input characters and convert them to integers if the functions cannot handle all possible inputs. For example, formatted input functions such as scanf(), fscanf(), vscanf(), and vfscanf() can be used to read string data from stdin or (in the cases of fscanf() and vfscanf()) other input streams. These functions work fine for valid integer values but lack robust error handling for invalid values.

Alternatively, input character data as a null-terminated byte string and convert to an integer value using strtol() or a related function. (See recommendation INT06-C. Use strtol() or a related function to convert a string token to an integer.)

Noncompliant Code Example

This noncompliant code example uses the scanf() function to read a string from stdin and convert it to a long. The scanf() and fscanf() functions have undefined behavior if the value of the result of this operation cannot be represented as an integer.

long sl;

if (scanf("%ld", &sl) != 1) {
  /* handle error */
}

In general, do not use scanf() to parse integers or floating-point numbers from input strings because the input could contain numbers not representable by the argument type.

Compliant Solution (Linux)

This compliant example uses the Linux scanf() implementation's built in error handling to validate input. On Linux platforms, scanf() sets errno to ERANGE if the result of integer conversion cannot be represented within the size specified by the format string [Linux 2008]. Note that this is a platform dependent solution. Therefore, this should only be used where portability is not a concern.

long sl;
errno = 0;

if (scanf("%ld", &sl) != 1) {
  /* handle error */
}
else if (ERANGE == errno) {
  if (puts("number out of range\n") == EOF) {
      /* Handle error */
  }
}

Compliant Solution

This compliant example uses fgets() to input a string and strtol() to convert the string to an integer. Error checking is provided to make sure that the value is a valid integer in the range of long.

char buff[25];
char *end_ptr;
long sl;

if (fgets(buff, sizeof(buff), stdin) == NULL) {
  if (puts("EOF or read error\n") == EOF) {
    /* Handle error */
  }
} else {
  errno = 0;

  sl = strtol(buff, &end_ptr, 10);

  if (ERANGE == errno) {
    if (puts("number out of range\n") == EOF) {
      /* Handle error */
    }
  }
  else if (end_ptr == buff) {
    if (puts("not valid numeric input\n") == EOF) {
      /* Handle error */
    }
  }
  else if ('\n' != *end_ptr && '\0' != *end_ptr) {
    if (puts("extra characters on input line\n") == EOF) {
      /* Handle error */
    }
  }
}

Note that this solution treats any trailing characters, including white-space characters, as an error condition.

Risk Assessment

While it is relatively rare for a violation of this recommendation to result in a security vulnerability, it can easily result in lost or misinterpreted data.

Automated Detection

Related Vulnerabilities

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

Related Guidelines

CERT C++ Secure Coding Standard: INT05-CPP. Do not use input functions to convert character data if they cannot handle all possible inputs

ISO/IEC 9899:1999 Section 7.20.1.4, "The strtol, strtoll, strtoul, and strtoull functions," and Section 7.19.6, "Formatted input/output functions"

MITRE CWE: CWE-192, "Integer Coercion Error"

MITRE CWE: CWE-197, "Numeric Truncation Error"

Bibliography

[Klein 2002]
[Linux 2008] scanf(3)

INT04-C. Enforce limits on integer values originating from untrusted sources      04. Integers (INT)      INT06-C. Use strtol() or a related function to convert a string token to an integer