FIO30-C. Exclude user input from format strings

Never call any formatted I/O function with a format string containing user input.

An attacker who can fully or partially control the contents of a format string, can crash a vulnerable process, view the contents of the stack, view memory content, or write to an arbitrary memory location and consequently execute arbitrary code with the permissions of the vulnerable process. See [[Seacord 05]] for a detailed explanation of how format string vulnerabilities may be exploited.

Formatted output functions are particularly dangerous because many programmers are unaware of their capabilities (for example, they can write an integer value to a specified address using the %n conversion specifier).

Non-Compliant Code Example

This non-compliant code example illustrates the incorrect_password() function which is called during identification and authentication if the specified user is not found or the password is incorrect to display an error message. The function accepts the name of the user as a null-terminated byte string referenced by user. This is an excellent example of data that originates from an untrusted, unauthenticated user. The function constructs an error message which is then output to stderr using the C99 standard fprintf() function.

void incorrect_password(const char *user) {
  /* user names are restricted to 256 characters or less */
  size_t len = strlen(user) + sizeof("%s could not be authenticated.") - 1;
  char *msg = (char *)malloc(len);
  if (!msg) {
    /* handle error condition */
  }
  snprintf(msg, len, "%s could not be authenticated.", user);
  fprintf(stderr, msg);
  free(msg);
}

The incorrect_password() function constructs msg in dynamically allocated memory by first calculating the size of the message, allocating dynamic storage, and then constructing the message in the allocated memory using the snprintf())) function. The addition operations are not checked for integer overflow, because the length of the string referenced by {{user is known to be have a length of 256 or less. Because the %s characters are replaced by the string referenced by {user in the call to snprintf(), one less byte is required to store the resulting string and terminating null byte character. This is a common idiom for displaying the same message in multiple locations or when the message is difficult to build. The resulting code contains a format string vulnerability, however, because the msg includes untrusted user input and is passed as the format string argument in the call to fprintf().

Compliant Solution

This compliant solution passes the untrusted user input as one of the variadic arguments to fprintf() and not as part of the format string; eliminating the possibility of a format string vulnerability.

void incorrect_password(const char *user) {
    fprintf(stderr, "%s could not be authenticated.", user);
  }
}

Non-Compliant Code Example (POSIX)

This non-compliant code example is exactly the same as the first non-compliant code example, but uses the POSIX function syslog() [[Open Group 04]] instead of the fprintf() function, which is also susceptible to format string vulnerabilities.

void incorrect_password(const char *user, const char *password) {
  /* user names are restricted to 256 characters or less */
  size_t len = strlen(user) + sizeof("%s could not be authenticated.") - 1;
  char *msg = (char *)malloc(len);
  if (!msg) {
    /* handle error condition */
  }
  snprintf(msg, len, "%s could not be authenticated.", user);
  syslog(LOG_INFO, msg);
  free(msg);
}

The syslog() function first appeared in BSD 4.2 and is supported by Linux and other modern Unix implementations. It is not available on Windows systems.

Compliant Solution (POSIX)

This compliant solution passes the untrusted user input as one of the variadic arguments to syslog() instead of including it in the format string.

void incorrect_password(const char *user) {
    syslog(LOG_INFO, "%s could not be authenticated.", user);
  }
}

Risk Assessment

Failing to exclude user input from format specifiers may allow an attacker to crash a vulnerable process, view the contents of the stack, view memory content, or write to an arbitrary memory location and consequently execute arbitrary code with the permissions of the vulnerable process.

Two recent examples of format string vulnerabilities resulting from a violation of this rule include Ettercap and Samba. In Ettercap v.NG-0.7.2, the ncurses user interface suffers from a format string defect. The curses_msg() function in ec_curses.c calls wdg_scroll_print(), which takes a format string and its parameters and passes it to vw_printw(). The curses_msg() function uses one of its parameters as the format string. This input can include user-data, allowing for a format string vulnerability [[VU#286468]]. The Samba AFS ACL mapping VFS plug-in fails to properly sanitize user-controlled filenames that are used in a format specifier supplied to snprintf(). This security flaw becomes exploitable when a user is able to write to a share that uses Samba's afsacl.so library for setting Windows NT access control lists on files residing on an AFS file system.

Related Vulnerabilities

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

References

[[ISO/IEC 9899-1999]] Section 7.19.6, "Formatted input/output functions"
[[MITRE 07]] CWE ID 134, "Uncontrolled Format String"
[[Open Group 04]] syslog()
[[Seacord 05]] Chapter 6, "Formatted Output"
[[Viega 05]] Section 5.2.23, "Format string problem"
[[VU#286468]]
[[VU#649732]]

FIO15-A. Avoid taking conditional actions based on path or file names      09. Input Output (FIO)       FIO31-C. Do not simultaneously open the same file multiple times