SIG30-C. Call only asynchronous-safe functions within signal handlers

Call only abort(), _Exit(), and signal() from within a signal handler in the C Standard Library. 

Section 7.14.1.1, para. 5, of the C standard [ISO/IEC 9899:2011], states that if the signal occurs other than as the result of calling the abort() or raise() function, the behavior is undefined if

the signal handler calls any function in the standard library other than the abort function, the _Exit function, or the signal function with the first argument equal to the signal number corresponding to the signal that caused the invocation of the handler. 

Many systems define an implementation-specific list of asynchronous-safe functions. These functions can also be called within a signal handler. This restriction applies to library functions as well as application-defined functions.

According to Section 7.14.1.1 of the C Rationale [ISO/IEC 2003],

When a signal occurs, the normal flow of control of a program is interrupted. If a signal occurs that is being trapped by a signal handler, that handler is invoked. When it is finished, execution continues at the point at which the signal occurred. This arrangement can cause problems if the signal handler invokes a library function that was being executed at the time of the signal.

In general, I/O functions are not safe to invoke inside signal handlers. Check your system's asynchronous-safe functions before using them in signal handlers.

Noncompliant Code Example

In this noncompliant code example, the program allocates a string on the heap and uses it to log messages in a loop. The program also registers the signal handler int_handler() to handle the terminal interrupt signal SIGINT. The int_handler() function logs the last message, calls free(), and exits.

#include <signal.h>
#include <stdio.h>
#include <stdlib.h>

enum { MAXLINE = 1024 };
char *info = NULL;

void log_message(void) {
  fprintf(stderr, info);
}

void handler(int signum) {
  log_message();
  free(info);
  info = NULL;
}

int main(void) {
  if (signal(SIGINT, handler) == SIG_ERR) {
    /* Handle error */
  }
  info = (char*)malloc(MAXLINE);
  if (info == NULL) {
    /* Handle Error */
  }

  while (1) {
    /* Main loop program code */

    log_message();

    /* More program code */
  }
  return 0;
}

This program's signal handler has four problems. The first is that it is unsafe to call the fprintf() function from within a signal handler because the handler may be called when global data (such as stderr) is in an inconsistent state. In general, it is not safe to invoke I/O functions within a signal handler.

The second problem is that the free() function is also not asynchronous-safe, and its invocation from within a signal handler is also a violation of this rule. If an interrupt signal is received during the free() call in handler(), the heap may be corrupted.

The third problem is that if SIGINT occurs after the call to free(), the memory referenced by info is freed twice. This is a violation of MEM31-C. Free dynamically allocated memory exactly once and SIG31-C. Do not access shared objects in signal handlers.

The fourth problem is that the signal handler reads the variable info, which is not declared to be of type volatile sig_atomic_t. This is a violation of SIG31-C. Do not access shared objects in signal handlers.

Furthermore, there are problems in the main() function as well, such as the possibility that the signal handler might get invoked during the call to malloc() in main().

Noncompliant Code Example

Invoking the longjmp() function from within a signal handler can lead to undefined behavior if it results in the invocation of any non-asynchronous-safe functions, likely compromising the integrity of the program. Consequently, neither longjmp() nor the POSIX siglongjmp() should ever be called from within a signal handler.

This noncompliant code example is similar to a vulnerability in an old version of Sendmail [VU #834865]. The intent is to execute code in a main() loop, which also logs some data. Upon receiving a SIGINT, the program transfers out of the loop, logs the error, and terminates.

However, an attacker can exploit this noncompliant code example by generating a SIGINT just before the second if statement in log_message(). This results in longjmp() transferring control back to main(), where log_message() is called again. However, the first if statement would not be executed this time (because buf is not set to NULL as a result of the interrupt), and the program would write to the invalid memory location referenced by buf0.

#include <setjmp.h>
#include <signal.h>
#include <stdlib.h>

enum { MAXLINE = 1024 };
static jmp_buf env;

void handler(int signum) {
  longjmp(env, 1);
}

void log_message(char *info1, char *info2) {
  static char *buf = NULL;
  static size_t bufsize;
  char buf0[MAXLINE];

  if (buf == NULL) {
    buf = buf0;
    bufsize = sizeof(buf0);
  }

  /*
   *  Try to fit a message into buf, else re-allocate
   *  it on the heap and then log the message.
   */

/*** VULNERABILITY IF SIGINT RAISED HERE ***/

  if (buf == buf0) {
    buf = NULL;
  }
}

int main(void) {
  if (signal(SIGINT, handler) == SIG_ERR) {
    /* Handle error */
  }
  char *info1;
  char *info2;

  /* info1 and info2 are set by user input here */

  if (setjmp(env) == 0) {
    while (1) {
      /* Main loop program code */
      log_message(info1, info2);
      /* More program code */
    }
  }
  else {
    log_message(info1, info2);
  }

  return 0;
}

Compliant Solution

In this compliant solution, the call to longjmp() is removed; the signal handler sets an error flag of type volatile sig_atomic_t instead.

#include <signal.h>
#include <stdlib.h>

enum { MAXLINE = 1024 };
volatile sig_atomic_t eflag = 0;

void handler(int signum) {
  eflag = 1;
}

void log_message(char *info1, char *info2) {
  static char *buf = NULL;
  static size_t bufsize;
  char buf0[MAXLINE];

  if (buf == NULL) {
    buf = buf0;
    bufsize = sizeof(buf0);
  }

  /*
   *  Try to fit a message into buf, else re-allocate
   *  it on the heap and then log the message.
   */
  if (buf == buf0) {
    buf = NULL;
  }
}

int main(void) {
  if (signal(SIGINT, handler) == SIG_ERR) {
    /* Handle error */
  }
  char *info1;
  char *info2;

  /* info1 and info2 are set by user input here */

  while (!eflag) {
    /* Main loop program code */
    log_message(info1, info2);
    /* More program code */
  }

  log_message(info1, info2);

  return 0;
}

Implementation Details

POSIX

The following table from the the Open Group Base Specifications [Open Group 2004] defines a set of functions that are asynchronous-signal-safe. Applications may invoke these functions, without restriction, from a signal handler.

Asynchronous-Signal-Safe Functions

All functions not listed in this table are considered to be unsafe with respect to signals. In the presence of signals, all functions defined by IEEE standard 1003.1-2001 behave as defined when called from or interrupted by a signal handler, with a single exception: when a signal interrupts an unsafe function and the signal handler calls an unsafe function, the behavior is undefined.

Note that although raise() is on the list of asynchronous-safe functions, it should not be called within a signal handler if the signal occurs as a result of abort() or raise() function.

Section 7.14.1.1, para. 4, of the C standard [ISO/IEC 9899:2011] states:

If the signal occurs as the result of calling the abort or raise function, the signal handler shall not call the raise function.

(See also undefined behavior 131 of Annex J.)

Noncompliant Code Example

In this noncompliant code example, the int_handler() function is used to carry out SIGINT-specific tasks and then raises a SIGTERM. However, there is a nested call to the raise() function, which results in undefined behavior.

void term_handler(int signum) {
  /* SIGTERM handling specific */
}
 
void int_handler(int signum) {
  /* SIGINT handling specific */
  if (raise(SIGTERM) != 0) {  /* violation */
    /* Handle error */
  }
}
 
int main(void) {
  if (signal(SIGTERM, term_handler) == SIG_ERR) {
    /* Handle error */
  }
  if (signal(SIGINT, int_handler) == SIG_ERR) {
    /* Handle error */
  }
 
  /* Program code */
  if (raise(SIGINT) != 0) {
    /* Handle error */
  }
  /* More code */
 
  return EXIT_SUCCESS;
}

Compliant Solution

In this compliant solution, the call to the raise() function inside handler() has been replaced by a direct call to log_msg().

#include <signal.h>

void log_msg(int signum) {
  /* Log error message in some asynchronous-safe manner */
}

void handler(int signum) {
  /* Do some handling specific to SIGINT */
  log_msg(SIGUSR1);
}

int main(void) {
  if (signal(SIGUSR1, log_msg) == SIG_ERR) {
    /* Handle error */
  }
  if (signal(SIGINT, handler) == SIG_ERR) {
    /* handle error */
  }

  /* program code */
  if (raise(SIGINT) != 0) {
    /* Handle error */
  }
  /* More code */

  return 0;
}

Compliant Solution (POSIX)

If a signal handler is assigned using the POSIX sigaction() function, the signal handler may safely call raise().

The POSIX standard is contradictory regarding raise() in signal handlers. The POSIX standard [Open Group 2004] prohibits signal handlers installed using signal() from calling the raise() function if the signal occurs as the result of calling the raise(), kill(), pthread_kill(), or sigqueue() functions. However, it also requires that the raise() function may be safely called within any signal handler. Consequently, it is not clear whether it is safe for POSIX applications to call raise() in signal handlers installed using signal(), but it is safe to call raise() in signal handlers installed using sigaction().

#include <signal.h>

void log_msg(int signum) {
  /* Log error message in some asynchronous-safe manner */
}

void handler(int signum) {
  /* Do some handling specific to SIGINT */
  if (raise(SIGUSR1) != 0) {
    /* Handle error */
  }
}

int main(void) {
  struct sigaction act;
  act.sa_flags = 0;
  if (sigemptyset(&act.sa_mask) != 0) {
    /* Handle error */
  }
  act.sa_handler = log_msg;
  if (sigaction(SIGUSR1, &act, NULL) != 0) {
    /* Handle error */
  }
  act.sa_handler = handler;
  if (sigaction(SIGINT, &act, NULL) != 0) {
    /* Handle error */
  }

  /* program code */
  if (raise(SIGINT) != 0) {
    /* Handle error */
  }
  /* More code */

  return 0;
}

POSIX recommends sigaction() and deprecates signal(). Unfortunately, sigaction() is not defined in the C standard and is consequently not as portable a solution.

OpenBSD

The OpenBSD signal() man page identifies functions that are asynchronous-signal safe. Applications may consequently invoke them, without restriction, from a signal handler.

The OpenBSD signal() manual page lists a few additional functions that are asynchronous-safe in OpenBSD but "probably not on other systems," including snprintf(), vsnprintf(), and syslog_r() (but only when the syslog_data struct is initialized as a local variable).

Compliant Solution

Signal handlers should be as concise as possible, ideally, unconditionally setting a flag and returning. They may also call the _Exit() function. Finally, they may call other functions provided that all implementations to which the code is ported guarantee that these functions are asynchronous-safe.

This example code achieves compliance with this rule by moving the final log message and call to free() outside the signal handler.

#include <signal.h>
#include <stdio.h>
#include <stdlib.h>

enum { MAXLINE = 1024 };
volatile sig_atomic_t eflag = 0;
char *info = NULL;

void log_message(void) {
  fprintf(stderr, info);
}

void handler(int signum) {
  eflag = 1;
}

int main(void) {
  if (signal(SIGINT, handler) == SIG_ERR) {
    /* Handle error */
  }
  info = (char*)malloc(MAXLINE);
  if (info == NULL) {
    /* Handle error */
  }

  while (!eflag) {
    /* Main loop program code */

    log_message();

    /* More program code */
  }

  log_message();
  free(info);
  info = NULL;

  return 0;
}

Risk Assessment

Invoking functions that are not asynchronous-safe from within a signal handler may result in privilege escalation and other attacks.

Automated Detection

Related Vulnerabilities

For an overview of software vulnerabilities resulting from improper signal handling, see Zalewski's paper on understanding, exploiting, and preventing signal-handling-related vulnerabilities [Zalewski 2001]. VU #834865 describes a vulnerability resulting from a violation of this rule.

Another notable case where using the longjmp() function in a signal handler caused a serious vulnerability is wu-ftpd 2.4 [Greenman 1997]. The effective user ID is set to zero in one signal handler. If a second signal interrupts the first, a call is made to longjmp(), returning the program to the main thread but without lowering the user's privileges. These escalated privileges can be used for further exploitation.

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

Related Guidelines

CERT C++ Secure Coding Standard: SIG30-CPP. Call only asynchronous-safe functions within signal handlers

ISO/IEC 9899:2011 Section 7.14, "Signal handling <signal.h>"

ISO/IEC TR 17961 (Draft) Calling functions in the C Standard Library other than abort, _Exit, and signal from within a signal handler [asyncsig]

MITRE CWE: CWE ID 479, "Unsafe function call from a signal handler"

Bibliography

[Dowd 2006] Chapter 13, "Synchronization and State"
[ISO/IEC 2003] Section 5.2.3, "Signals and interrupts"
[Open Group 2004] longjmp()
[OpenBSD] signal() Man Page
[Zalewski 2001]