STR06-A. Do not assume that strtok() leaves the parse string unchanged

The C99 [[ISO/IEC 9899-1999]] function strtok() is a string tokenization function that takes three arguments: an initial string to be parsed, a const-qualified character delimiter, and a pointer to a pointer to modify to return the result.

The first time strtok() is called, the string is parsed into tokens, character delimiter, and address of the variable in which to return the result are passed as arguments. The strtok() function parses the string up to the first instance of the delimiter character, replaces the character in place with a null byte ('\0'), and puts the address of the first character in the token to the passed-in variable. Subsequent calls to strtok() begin parsing immediately after the most recently-placed null character.

Because strtok() modifies its argument, the string is subsequently unsafe and cannot be used in its original form. If you need to preserve the original string, copy it into a buffer and pass the address of the buffer to strtok() instead of the original string.

Non-Compliant Code Example

In this example, the strtok() function is used to parse the first argument into colon-delimited tokens; it outputs each word from the string on a new line. Assume that PATH is "/usr/bin:/usr/sbin:/sbin".

char *token;
char *path = getenv("PATH");

token = strtok(path, ":");
puts(token);

while (token = strtok(0, ":")) {
  puts(token);
}

printf("PATH: %s\n", path);
/* PATH is now just "/usr/bin" */

After the loop ends, path is modified as follows: "/usr/bin\0/bin\0/usr/sbin\0/sbin\0". This is an issue because the local path variable becomes /usr/bin and because the environment variable PATH has been unintentionally changed, which could have unintended consequences.

Compliant Solution

In this solution the string being tokenized is copied into a temporary buffer which is not referenced after the calls to strtok():

char *token;
char *path = getenv("PATH");
/* PATH is something like "/usr/bin:/bin:/usr/sbin:/sbin" */

char *copy = (char *)malloc(strlen(path) + 1);
if (copy == NULL) {
  /* handle error */
}
strcpy(copy, path);
token = strtok(copy, ":");
puts(token);

while (token = strtok(0, ":")) {
  puts(token);
}

printf("PATH: %s\n", path);
/* PATH is still "/usr/bin:/bin:/usr/sbin:/sbin" */

Another possibility is to provide your own implementation of strtok() that does not modify the initial arguments.

Risk Assessment

To quote the Linux Programmer's Manual (man) page on strtok(3):

Never use this function. This function modifies its first argument. The identity of the delimiting character is lost. This function cannot be used on constant strings.

However, improper strtok() use will probably only result in truncated data, producing unexpected results later in program execution.

h3 Automated Detection

Fortify SCA Version 5.0 is able to detect violations of this recommendation.

Related Vulnerabilities

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

References

[[ISO/IEC 9899-1999]] Section 7.21.5.8, "The strtok function"
[Unix Man page] strtok(3)

STR05-A. Prefer making string literals const-qualified      07. Characters and Strings (STR)       STR07-A. Use plain char for character data