ENV31-C. Do not rely on an environment pointer following an operation that may invalidate it

Some environments provide environment pointers that are valid when main() is called, but may be invalided by operations that modify the environment.

According to C99: [[ISO/IEC 9899:1999]]

In a hosted environment, the main function receives a third argument, char *envp[], that points to a null-terminated array of pointers to char, each of which points to a string that provides information about the environment for this execution of the program.

Consequently, under a hosted environments it is possible to access the environment through a modified form of main():

main(int argc, char *argv[], char *envp[])

However, modifying the environment by using the setenv() or putenv() functions, or by any other means, may cause the environment memory to be reallocated, with the result that envp now references an incorrect location. For example, POSIX says the following: [[Open Group 04]]

Unanticipated results may occur if setenv() changes the external variable environ.  In particular, if the optional envp argument to main() is present, it is not changed, and as a result may point to an obsolete copy of the environment (as may any other copy of environ).

Microsoft notes the following about getenv(): [[MSDN]]

getenv and _putenv use the copy of the environment pointed to by the global variable _environ to access the environment. getenv operates only on the data structures accessible to the runtime library and not on the environment "segment" created for the process by the operating system. Consequently, programs that use the envp argument to main or wmain may retrieve invalid information.

The Visual C++ reference notes the following about envp: [[MSDN]]

The environment block passed to main and wmain is a "frozen" copy of the current environment. If you subsequently change the environment via a call to putenv or _wputenv, the current environment (as returned by getenv/_wgetenv and the _environ/ _wenviron variable) will change, but the block pointed to by envp will not change.

When compiled with GCC version 3.4.6 and run on a 32-bit Intel GNU/Linux machine, the following code:

extern char **environ;

/* ... */

int main(int argc, char const *argv[], char const *envp[]) {
   printf("environ:  %p\n", environ);
   printf("envp:     %p\n", envp);
   setenv("MY_NEW_VAR", "new_value", 1);
   puts("--Added MY_NEW_VAR--");
   printf("environ:  %p\n", environ);
   printf("envp:     %p\n", envp);
}

Yields:

% ./envp-environ
environ: 0xbf8656ec
envp:    0xbf8656ec
--Added MY_NEW_VAR--
environ: 0x804a008
envp:    0xbf8656ec

It is evident from these results that the environment has been relocated as a result of the call to setenv().

Non-Compliant Code Example (POSIX)

After a call to the POSIX setenv() function, or other function that modifies the environment, the envp pointer may no longer reference the environment.

int main(int argc, char const *argv[], char const *envp[]) {
   size_t i;
   if (setenv("MY_NEW_VAR", "new_value", 1) != 0) {
     /* Handle Error */
   }
   if (envp != NULL) {
      for (i = 0; envp[i] != NULL; i++) {
         puts(envp[i]);
      }
   }
   return 0;
}

Because envp no longer points to the current environment, this program has undefined behavior.

Compliant Solution (POSIX)

Use environ in place of envp when defined.

extern char **environ;

/* ... */

int main(int argc, char const *argv[]) {
   size_t i;
   if (setenv("MY_NEW_VAR", "new_value", 1) != 0) {
     /* Handle Error */
   }
   if (environ != NULL) {
      for (i = 0; environ[i] != NULL; i++) {
         puts(environ[i]);
      }
   }
   return 0;
}

Non-Compliant Code Example (Windows)

After a call to the Windows _putenv_s() function, or other function that modifies the environment, the envp pointer may no longer reference the environment.

int main(int argc, char const *argv[], char const *envp[]) {
   size_t i;
   if (_putenv_s("MY_NEW_VAR", "new_value", 1) != 0) {
     /* Handle Error */
   }
   if (envp != NULL) {
      for (i = 0; envp[i] != NULL; i++) {
         puts(envp[i]);
      }
   }
   return 0;
}

Because envp no longer points to the current environment, this program has undefined behavior.

Compliant Solution (Windows)

Use _environ in place of envp when defined.

_CRTIMP extern char **_environ;

/* ... */

int main(int argc, char const *argv[]) {
   size_t i;
   if (_putenv_s("MY_NEW_VAR", "new_value", 1) != 0) {
     /* Handle Error */
   }
   if (_environ != NULL) {
      for (i = 0; _environ[i] != NULL; i++) {
         puts(_environ[i]);
      }
   }
   return 0;
}

Compliant Solution

Note: if you have a great deal of unsafe envp code, you can save time in your remediation by aliasing. Change:

int main(int argc, char *argv[], char *envp[]) {
  /* ... */
}

To:

#if  defined (_POSIX_) || defined (__USE_POSIX)
  extern char **environ;
  #define envp environ
#else
  _CRTIMP extern char **_environ;
  #define envp _environ
#endif

int main(int argc, char *argv[]) {
  /* ... */
}

Risk Assessment

Using the envp environment pointer after the environment has been modified may result in undefined behavior.

Related Vulnerabilities

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

References

[[ISO/IEC 9899:1999]] Section J.5.1, "Environment Arguments"
[[MSDN]] getenv, _wgetenv, _environ, _wenviron, _putenv_s, _wputenv_s
[[Open Group 04]] setenv()

ENV30-C. Do not modify the string returned by getenv()      10. Environment (ENV)       ENV32-C. No atexit handler should terminate in any way other than by returning