1SIGNAL(2) Linux Programmer's Manual SIGNAL(2)
2
6 signal - ANSI C signal handling
7
9 #include <signal.h>
10 typedef void (*sighandler_t)(int);
12 sighandler_t signal(int signum, sighandler_t handler);
14
16 The behavior of signal() varies across UNIX versions, and has also var‐
17 ied historically across different versions of Linux. Avoid its use:
18 use sigaction(2) instead. See Portability below.
19 signal() sets the disposition of the signal signum to handler, which is
21 either SIG_IGN, SIG_DFL, or the address of a programmer-defined func‐
22 tion (a "signal handler").
23 If the signal signum is delivered to the process, then one of the fol‐
25 lowing happens:
26 * If the disposition is set to SIG_IGN, then the signal is ignored.
28 * If the disposition is set to SIG_DFL, then the default action asso‐
30 ciated with the signal (see signal(7)) occurs.
31 * If the disposition is set to a function, then first either the dis‐
33 position is reset to SIG_DFL, or the signal is blocked (see Porta‐
34 bility below), and then handler is called with argument signum. If
35 invocation of the handler caused the signal to be blocked, then the
36 signal is unblocked upon return from the handler.
37 The signals SIGKILL and SIGSTOP cannot be caught or ignored.
39
41 signal() returns the previous value of the signal handler, or SIG_ERR
42 on error. In the event of an error, errno is set to indicate the
43 cause.
44
46 EINVAL signum is invalid.
47
49 POSIX.1-2001, POSIX.1-2008, C89, C99.
50
52 The effects of signal() in a multithreaded process are unspecified.
53 According to POSIX, the behavior of a process is undefined after it
55 ignores a SIGFPE, SIGILL, or SIGSEGV signal that was not generated by
56 kill(2) or raise(3). Integer division by zero has undefined result.
57 On some architectures it will generate a SIGFPE signal. (Also dividing
58 the most negative integer by -1 may generate SIGFPE.) Ignoring this
59 signal might lead to an endless loop.
60 See sigaction(2) for details on what happens when the disposition
62 SIGCHLD is set to SIG_IGN.
63 See signal-safety(7) for a list of the async-signal-safe functions that
65 can be safely called from inside a signal handler.
66 The use of sighandler_t is a GNU extension, exposed if _GNU_SOURCE is
68 defined; glibc also defines (the BSD-derived) sig_t if _BSD_SOURCE
69 (glibc 2.19 and earlier) or _DEFAULT_SOURCE (glibc 2.19 and later) is
70 defined. Without use of such a type, the declaration of signal() is
71 the somewhat harder to read:
72 void ( *signal(int signum, void (*handler)(int)) ) (int);
74 Portability
76 The only portable use of signal() is to set a signal's disposition to
77 SIG_DFL or SIG_IGN. The semantics when using signal() to establish a
78 signal handler vary across systems (and POSIX.1 explicitly permits this
79 variation); do not use it for this purpose.
80 POSIX.1 solved the portability mess by specifying sigaction(2), which
82 provides explicit control of the semantics when a signal handler is
83 invoked; use that interface instead of signal().
84 In the original UNIX systems, when a handler that was established using
86 signal() was invoked by the delivery of a signal, the disposition of
87 the signal would be reset to SIG_DFL, and the system did not block
88 delivery of further instances of the signal. This is equivalent to
89 calling sigaction(2) with the following flags:
90 sa.sa_flags = SA_RESETHAND | SA_NODEFER;
92 System V also provides these semantics for signal(). This was bad
94 because the signal might be delivered again before the handler had a
95 chance to reestablish itself. Furthermore, rapid deliveries of the
96 same signal could result in recursive invocations of the handler.
97 BSD improved on this situation, but unfortunately also changed the
99 semantics of the existing signal() interface while doing so. On BSD,
100 when a signal handler is invoked, the signal disposition is not reset,
101 and further instances of the signal are blocked from being delivered
102 while the handler is executing. Furthermore, certain blocking system
103 calls are automatically restarted if interrupted by a signal handler
104 (see signal(7)). The BSD semantics are equivalent to calling sigac‐
105 tion(2) with the following flags:
106 sa.sa_flags = SA_RESTART;
108 The situation on Linux is as follows:
110 * The kernel's signal() system call provides System V semantics.
112 * By default, in glibc 2 and later, the signal() wrapper function does
114 not invoke the kernel system call. Instead, it calls sigaction(2)
115 using flags that supply BSD semantics. This default behavior is pro‐
116 vided as long as a suitable feature test macro is defined:
117 _BSD_SOURCE on glibc 2.19 and earlier or _DEFAULT_SOURCE in glibc
118 2.19 and later. (By default, these macros are defined; see fea‐
119 ture_test_macros(7) for details.) If such a feature test macro is
120 not defined, then signal() provides System V semantics.
121
123 kill(1), alarm(2), kill(2), pause(2), sigaction(2), signalfd(2), sig‐
124 pending(2), sigprocmask(2), sigsuspend(2), bsd_signal(3), killpg(3),
125 raise(3), siginterrupt(3), sigqueue(3), sigsetops(3), sigvec(3),
126 sysv_signal(3), signal(7)
127
129 This page is part of release 5.02 of the Linux man-pages project. A
130 description of the project, information about reporting bugs, and the
131 latest version of this page, can be found at
132 https://www.kernel.org/doc/man-pages/.
133Linux 2017-09-15 SIGNAL(2)