NAME

sigaction — software signal facilities

SYNOPSIS

#include <signal.h>

struct sigaction {
	union {		/* signal handler */
		void	(*__sa_handler)(int);
		void	(*__sa_sigaction)(int, siginfo_t *, void *);
	} __sigaction_u;
	sigset_t sa_mask;          /* signal mask to apply */
	int	 sa_flags;         /* see signal options below */
};

#define sa_handler __sigaction_u.__sa_handler
#define sa_sigaction __sigaction_u.__sa_sigaction

int
sigaction(int sig, const struct sigaction *act, struct sigaction *oact);

DESCRIPTION

The system defines a set of signals that may be delivered to a process. Signal delivery resembles the occurrence of a hardware interrupt: the signal is normally blocked from further occurrence, the current process context is saved, and a new one is built. A process may specify a handler to which a signal is delivered, or specify that a signal is to be ignored. A process may also specify that a default action is to be taken by the system when a signal occurs. A signal may also be blocked, in which case its delivery is postponed until it is unblocked. The action to be taken on delivery is determined at the time of delivery. Normally, signal handlers execute on the current stack of the process. This may be changed, on a per-handler basis, so that signals are taken on a special signal stack.

Signal routines normally execute with the signal that caused their invocation blocked, but other signals may yet occur. A global signal mask defines the set of signals currently blocked from delivery to a process. The signal mask for a process is initialized from that of its parent (normally empty). It may be changed with a sigprocmask(2) call, or when a signal is delivered to the process.

When a signal condition arises for a process, the signal is added to a set of signals pending for the process. If the signal is not currently blocked by the process then it is delivered to the process. Signals may be delivered any time a process enters the operating system (e.g., during a system call, page fault or trap, or clock interrupt). If multiple signals are ready to be delivered at the same time, any signals that could be caused by traps are delivered first. Additional signals may be processed at the same time, with each appearing to interrupt the handlers for the previous signals before their first instructions. The set of pending signals is returned by the sigpending(2) function. When a caught signal is delivered, the current state of the process is saved, a new signal mask is calculated (as described below), and the signal handler is invoked. The call to the handler is arranged so that if the signal handling routine returns normally the process will resume execution in the context from before the signal's delivery. If the process wishes to resume in a different context, then it must arrange to restore the previous context itself.

When a signal is delivered to a process, a new signal mask is installed for the duration of the process' signal handler (or until a sigprocmask(2) call is made). This mask is formed by taking the union of the current signal mask set, the signal to be delivered, and the signal mask sa_mask associated with the handler to be invoked, but always excluding SIGKILL and SIGSTOP.

sigaction() assigns an action for a signal specified by sig. If act is non-zero, it specifies an action (SIG_DFL, SIG_IGN, or a handler routine) and mask to be used when delivering the specified signal. If oact is non-zero, the previous handling information for the signal is returned to the user.

Once a signal handler is installed, it normally remains installed until another sigaction() call is made, or an execve(2) is performed. The value of sa_handler (or, if the SA_SIGINFO flag is set, the value of sa_sigaction instead) indicates what action should be performed when a signal arrives. A signal-specific default action may be reset by setting sa_handler to SIG_DFL. Alternately, if the SA_RESETHAND flag is set the default action will be reinstated when the signal is first posted. The defaults are process termination, possibly with core dump; no action; stopping the process; or continuing the process. See the signal list below for each signal's default action. If sa_handler is SIG_DFL, the default action for the signal is to discard the signal, and if a signal is pending, the pending signal is discarded even if the signal is masked. If sa_handler is set to SIG_IGN, current and pending instances of the signal are ignored and discarded. If sig is SIGCHLD and sa_handler is set to SIG_IGN, the SA_NOCLDWAIT flag (described below) is implied.

The signal mask sa_mask is typically manipulated using the sigaddset(3) family of functions.

Options may be specified by setting sa_flags. The meaning of the various bits is as follows:

After a fork(2) or vfork(2), all signals, the signal mask, the signal stack, and the restart/interrupt flags are inherited by the child.

execve(2) reinstates the default action for SIGCHLD and all signals which were caught; all other signals remain ignored. All signals are reset to be caught on the user stack and the signal mask remains the same; signals that restart pending system calls continue to do so.

The following is a list of all signals with names as in the include file <signal.h>:

RETURN VALUES

Upon successful completion, the value 0 is returned; otherwise the value -1 is returned and the global variable errno is set to indicate the error.

EXAMPLES

The handler routine can be declared:

void
handler(int sig)

void
handler(int sig, siginfo_t *sip, void *ctx)

Here sig is the signal number, into which the hardware faults and traps are mapped. If the SA_SIGINFO option is set, sip is a pointer to a siginfo_t as described in <sys/siginfo.h>. If SA_SIGINFO is not set, this pointer will be NULL instead. The function specified in sa_sigaction will be called instead of the function specified by sa_handler (note that in some implementations these are in fact the same). ctx may be cast to a pointer to ucontext_t which can be used to restore the thread's context from before the signal. On OpenBSD, ucontext_t is an alias for the sigcontext structure defined in <signal.h>. The contents of this structure are machine-dependent.

ERRORS

sigaction() will fail and no new signal handler will be installed if one of the following occurs:

[EFAULT]

Either act or oact points to memory that is not a valid part of the process address space.

[EINVAL]

sig is not a valid signal number.

[EINVAL]

An attempt is made to ignore or supply a handler for SIGKILL or SIGSTOP.

SEE ALSO

kill(1), kill(2), ptrace(2), sigaltstack(2), sigprocmask(2), sigsuspend(2), wait(2), setjmp(3), sigaddset(3), sigblock(3), sigpause(3), sigvec(3), tty(4)

STANDARDS

The sigaction() function conforms to IEEE Std 1003.1-2008 (“POSIX.1”).

The SA_ONSTACK flag and the SIGPROF, SIGSYS, SIGTRAP, SIGVTALRM, SIGXCPU, and SIGXFSZ signals conform to the X/Open System Interfaces option of that standard. The standard marks SIGPROF as obsolescent. The signals SIGEMT, SIGINFO, SIGIO, and SIGWINCH are Berkeley extensions. These signals are available on most BSD-derived systems. The SIGTHR signal is specific to OpenBSD and is part of the implementation of thread cancellation; sigaction and other signal interfaces may reject attempts to use or alter the handling of SIGTHR.

Signal handlers should be as minimal as possible, and use only signal-safe operations. The safest handlers only change a single variable of type volatile sig_atomic_t, which is inspected by an event loop. Other variables accessed inside the handler must be either const, or local to the handler. More complicated global variables (such as strings, structs, or lists) will require external methods to guarantee consistency, such as signal-blocking with sigprocmask(2).

More complicated handlers must restrict themselves to calling only the following list of signal-safe functions directly. Avoid abstracting the work to helper functions which are also called from other contexts because future coders will forget the signal-safe requirement.

Standard Interfaces:

_exit(), _Exit(), abort(), accept(), access(), alarm(), bind(), cfgetispeed(), cfgetospeed(), cfsetispeed(), cfsetospeed(), chdir(), chmod(), chown(), clock_gettime(), close(), connect(), creat(), dup(), dup2(), execl(), execle(), execv(), execve(), faccessat(), fchdir(), fchmod(), fchmodat(), fchown(), fchownat(), fcntl(), fdatasync(), fork(), fpathconf(), fstat(), fstatat(), fsync(), ftruncate(), futimens(), futimes(), getegid(), geteuid(), getgid(), getgroups(), getpeername(), getpgrp(), getpid(), getppid(), getsockname(), getsockopt(), getuid(), kill(), link(), linkat(), listen(), lseek(), lstat(), mkdir(), mkdirat(), mkfifo(), mkfifoat(), mknod(), mknodat(), open(), openat(), pathconf(), pause(), pipe(), poll(), pselect(), pthread_sigmask(), raise(), read(), readlink(), readlinkat(), recv(), recvfrom(), recvmsg(), rename(), renameat(), rmdir(), select(), send(), sendmsg(), sendto(), setgid(), setpgid(), setsid(), setsockopt(), setuid(), shutdown(), sigaction(), sigaddset(), sigdelset(), sigemptyset(), sigfillset(), sigismember(), signal(), sigpause(), sigpending(), sigprocmask(), sigsuspend(), sleep(), sockatmark(), socket(), socketpair(), stat(), strcat(), strcpy(), strncat(), strncpy(), symlink(), symlinkat(), sysconf(), tcdrain(), tcflow(), tcflush(), tcgetattr(), tcgetpgrp(), tcsendbreak(), tcsetattr(), tcsetpgrp(), time(), times(), umask(), uname(), unlink(), unlinkat(), utime(), utimensat(), utimes(), wait(), waitpid(), write(), and perhaps some others.

Extension Interfaces:

accept4(), chflags(), chflagsat(), dup3(), fchflags(), getentropy(), getresgid(), getresuid(), pipe2(), ppoll(), sendsyslog(), setresgid(), setresuid(), strlcat(), strlcpy(), wait3(), wait4().

Since signal-safe functions can encounter system call errors, errno should be protected inside the handler with the following pattern:

void
handler(int sig)
{
	int save_errno = errno;

	...
	errno = save_errno;
}

On OpenBSD, a few more functions are signal-safe (except when the format string contains floating-point arguments). These functions are expected to be unsafe on other systems, so be very cautious of the portability trap!