Page History

Many library functions accept a string or wide string argument with the constraint that the string they receive is properly null-terminated. Passing a character sequence or wide character sequence that is not null-terminated to such a function can result in accessing memory that is outside the bounds of the object. Do not pass a character sequence or wide character sequence that is not null-terminated to a library function that expects a string or wide string argument. 

Noncompliant Code Example

This code example is noncompliant because the character sequence c_str will not be null-terminated when passed as an argument to printf(). (See STR11-C. Do not specify the bound of a character array initialized with a string literal on how to properly initialize character arrays.)

#include <stdio.h>
 
void func(void) {
  char c_str[3] = "abc";
  printf("%s\n", c_str);
}

Compliant Solution

This compliant solution does not specify the bound of the character array in the array declaration. If the array bound is omitted, the compiler allocates sufficient storage to store the entire string literal, including the terminating null character.

#include <stdio.h>
 

Strings must contain a null-termination character at or before the address of the last element of the array before they can be safely passed as arguments to standard string-handling functions, such as strcpy() or strlen(). These functions, as well as other string-handling functions defined by the C Standard, depend on the existence of the string's null-termination character to determine the length of the string. Similarly, strings must be null-terminated before iterating on a character array where the termination condition of the loop depends on the existence of a null-termination character within the memory allocated for the string, as in the following example:

void func(void) {
  char ntbsc_str[16];

  for (size_t i = 0; i < sizeof(ntbs); ++i) {
    if (ntbs[i] == '\0') {
      break;
    }
  }

Failure to properly terminate strings can result in buffer overflows and other undefined behavior.

] = "abc";
  printf("%s\n", c_str);
}

Noncompliant Code Example

This code example is noncompliant because the wide character sequence cur_msg will not be null-terminated when passed to wcslen(). This will occur if lessen_memory_usage() is invoked while cur_msg_size still has its initial value of 1024.

#include <stdlib.h>
#include <wchar.h>
 
wchar_t *cur_msg = NULL;
size_t cur_msg_size = 1024;
size_t cur_msg_len = 0;

void lessen_memory_usage(void) {
  wchar_t *temp;
  size_t temp_size;

  /* ... */

  if (cur_msg != NULL) {
    temp_size = cur_msg_size / 2 + 1;
    temp = realloc(cur_msg, temp_size * sizeof(wchar_t));
    /* temp &and cur_msg may no longer be null-terminated */
    if (temp == NULL) {
      /* Handle error */
    }

    cur_msg = temp;
    cur_msg_size = temp_size;
    cur_msg_len = wcslen(cur_msg); 
  }
}

Compliant Solution

In this compliant solution, cur_msg will always be null-terminated when passed to wcslen():

#include <stdlib.h>
#include <wchar.h>
 
wchar_t *cur_msg = NULL;
size_t cur_msg_size = 1024;
size_t cur_msg_len = 0;

void lessen_memory_usage(void) {
  wchar_t *temp;
  size_t temp_size;

  /* ... */

  if (cur_msg != NULL) {
    temp_size = cur_msg_size / 2 + 1;
    temp = realloc(cur_msg, temp_size * sizeof(wchar_t));
    /* temp and cur_msg may no longer be null-terminated */
    if (temp == NULL) {
      /* Handle error */
    }

    cur_msg = temp;
    /* Properly null-terminate cur_msg */
    cur_msg[temp_size - 1] = L'\0'; 
    cur_msg_size = temp_size;
    cur_msg_len = wcslen(cur_msg); 
  }
}

Noncompliant Code Example (strncpy())

The Although the strncpy() function takes a string as input, it does not guarantee that the resulting value is still null-terminated. If In the following noncompliant code example, if no null character is contained in the first n characters of the source array, the result will not be null-terminated. Passing a non-null-terminated character sequence to strlen() results in is undefined behavior, as shown by this noncompliant code example:.

#include <string.h>
 
enum { NTBSSTR_SIZE = 32 };
 
size_t func(const char *source) {
  char ntbsc_str[NTBSSTR_SIZE];
  size_t ret = 0;

  ntbsif (source) {
    c_str[sizeof(ntbsc_str) - 1] = '\0';
    strncpy(ntbsc_str, source, sizeof(ntbsc_str));
    ret return= strlen(ntbsc_str);
}

Compliant Solution (Truncation)

The correct solution depends on the programmer's intent. If the intent is to truncate a string, this solution can be used:

#include <string.h>
 
enum { NTBS_SIZE = 32 };
 
size_t func(const char *source) else {
  char ntbs[NTBS_SIZE];

  strncpy(ntbs, source, sizeof(ntbs) - 1);
  ntbs[sizeof(ntbs) - 1] = '\0'; /* Handle null pointer */
  }
  return strlen( ntbs)ret;
}

Compliant Solution (

...

Truncation)

If the This compliant solution is correct if the programmer's intent is to copy without truncation, this example copies the data and guarantees that the resulting string is null-terminated. If the string cannot be copied, it is handled as an error condition.truncate the string:

#include <string.h>
 
enum { NTBSSTR_SIZE = 32 };
 
size_t func(const char *source) {
  char ntbsc_str[NTBSSTR_SIZE];
  size_t ret = 0;

  if (source) {
    if (strlen(source) <strncpy(c_str, source, sizeof(ntbs)) {c_str) - 1);
      strcpy(ntbs, source);
    } else {c_str[sizeof(c_str) - 1] = '\0';
    ret  /* Handle string-too-large */
    }= strlen(c_str);
  } else {
    /* Handle null pointer */
  }
  return strlen(ntbs)ret;
}

Compliant Solution (Truncation, strncpy_s())

The C Standard,

...

Annex K

...

strncpy_s() function can also be used to copy with truncation. The strncpy_s() function copies up to n characters from the source array to a destination array. If no null character was copied from the source array, then the nth position in the destination array is set to a null character, guaranteeing that the resulting string is null-terminated.

#define __STDC_WANT_LIB_EXT1__ 1
#include <string.h>
  
enum { NTBSSTR_SIZE = 32 };
  
size_t func(const char *source) {
  char ac_str[NTBSSTR_SIZE];
  size_t ret = 0;

  if (source) {
    errno_t err = strncpy_s(a
      c_str, sizeof(a)c_str), source, strnlen(source, 5sizeof(c_str))
    );
    if (err != 0) {
      /* Handle error */
    }
  } else {
    /* Handle nullret pointer */
  }
  return strlen_s(s, sizeof(a));
}

Noncompliant Code Example (realloc())

One method to decrease memory usage in critical situations when all available memory has been exhausted is to use the realloc() function to halve the size of message strings. The standard realloc() function has no concept of strings. As a result, if realloc() is called to decrease the memory allocated for a string, the null-termination character may be truncated.

This noncompliant code example fails to ensure that cur_msg is properly null-terminated:

#include <stdlib.h>
 
char *cur_msg = NULL;
size_t cur_msg_size = 1024;
 
void lessen_memory_usage(void) {
  char *temp;
  size_t temp_size;

  if (cur_msg != NULL) {
    temp_size = cur_msg_size / 2 + 1= strnlen(c_str, sizeof(c_str));
    temp = realloc(cur_msg, temp_size);}
    if (temp == NULL) } else {
      /* Handle null errorpointer */
    }
    cur_msg = temp;
    cur_msg_size = temp_size;
    fputs(stderr, cur_msg);
  }
}

Because realloc() does not guarantee that the string is properly null-terminated, and the function subsequently passes cur_msg to a library function (fputs()) that expects null-termination, the result is undefined behavior.

Compliant Solution (realloc())

return ret;
}

Compliant Solution (Copy without Truncation)

If the programmer's intent is to copy without truncation, this compliant solution copies the data and guarantees that the resulting array is null-terminated. If the string cannot be copied, it is handled as an error condition.In this compliant solution, the lessen_memory_usage() function ensures that the resulting string is always properly null-terminated:

#include <stdlib<string.h>
 
char *cur_msgenum { STR_SIZE = 32 NULL};
 
size_t cur_msg_size = 1024;

void lessen_memory_usage(voidfunc(const char *source) {
  char *tempc_str[STR_SIZE];
  size_t temp_size ret = 0;

  if (cur_msg != NULLsource) {
    temp_size = cur_msg_size / 2 + 1;if (strnlen(source, sizeof(c_str)) < sizeof(c_str)) {
    temp = reallocstrcpy(curc_msgstr, temp_sizesource);
    if  (tempret == NULL) strlen(c_str);
    } else {
      /* Handle errorstring-too-large */
    }
  }  cur_msg = temp;
    cur_msg_size = temp_size;

    else {
    /* EnsureHandle string is null-terminatednull pointer */
  }
  cur_msg[cur_msg_size - 1] = '\0';
  }
  fputs(stderr, cur_msg);
}return ret;
}

Note that this code is not bulletproof. It gracefully handles the case where source  is NULL, when it is a valid string, and when source is not null-terminated, but at least the first 32 bytes are valid. However, in cases where source is not NULL, but points to invalid memory, or any of the first 32 bytes are invalid memory, the first call to strnlen() will access this invalid memory, and the resulting behavior is undefined. Unfortunately, standard C provides no way to prevent or even detect this condition without some external knowledge about the memory source points to.

Risk Assessment

Failure to properly null-terminate a character sequence that is passed to a library function that expects a string can result in buffer overflows and the execution of arbitrary code with the permissions of the vulnerable process. Null-termination errors can also result in unintended information disclosure.

Automated Detection

Related Vulnerabilities

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

Related Guidelines

Key here (explains table format and definitions)

CERT-CWE Mapping Notes

Key here for mapping notes

CWE-119 and STR32-C

Independent( ARR30-C, ARR38-C, ARR32-C, INT30-C, INT31-C, EXP39-C, EXP33-C, FIO37-C) STR31-C = Subset( Union( ARR30-C, ARR38-C)) STR32-C = Subset( ARR38-C)

CWE-119 = Union( STR32-C, list) where list =

• Out-of-bounds reads or writes that do not involve non-null-terminated byte strings.

CWE-125 and STR32-C

Independent( ARR30-C, ARR38-C, EXP39-C, INT30-C) STR31-C = Subset( Union( ARR30-C, ARR38-C)) STR32-C = Subset( ARR38-C)

CWE-125 = Union( STR32-C, list) where list =

• Out-of-bounds reads that do not involve non-null-terminated byte strings.

CWE-123 and STR32-C

Independent(ARR30-C, ARR38-C) STR31-C = Subset( Union( ARR30-C, ARR38-C)) STR32-C = Subset( ARR38-C)

Intersection( CWE-123, STR32-C) =

• Buffer overflow from passing a non-null-terminated byte string to a standard C library copying function that expects null termination, and that overwrites an (unrelated) pointer

STR32-C - CWE-123 =

• Buffer overflow from passing a non-null-terminated byte string to a standard C library copying function that expects null termination, but it does not overwrite an (unrelated) pointer

CWE-123 – STR31-C =

• Arbitrary writes that do not involve standard C library copying functions, such as strcpy()

Bibliography

...

...

Image Modified Image Modified Image Modified