sigmask

Sigmask is a macro found in some Unix-like systems that produces a bit mask for a given signal number. It is part of the older signal handling interface and is mainly encountered in legacy code or systems that provide compatibility with early Unix APIs. The concept behind sigmask is to map a signal’s numeric identifier to a single bit in a process-wide mask used to control delivery of signals.

In many implementations, sigmask is defined in signal.h as a bit-shift expression, typically something like a macro defined as #define sigmask(sig) (1L << ((sig) - 1)). The exact type (for example long or unsigned long) and whether the macro exists can vary by platform. The macro is intended to be used with older facilities that manipulate a simple integer mask rather than a modern signal set object.

Historically, programs constructed a mask for a set of signals using sigmask and then used functions such as sigblock, sigsetmask, or sigpause to block, replace, or wait for signals. These primitives operate on a single integer mask representing blocked signals, in contrast to the POSIX approach of manipulating sets of signals.

Modern practice favors POSIX signal handling, which uses sigset_t objects and functions like sigemptyset, sigaddset, and sigprocmask (or pthread_sigmask for threads) to manipulate the signal mask in a portable and robust way. In contemporary code sigmask is generally deprecated or retained only for compatibility with legacy code.

If sigmask appears in code today, it is usually advisable to rewrite using sigset_t and sigprocmask, though some legacy interfaces may still rely on a numeric mask via sigblock or sigsetmask. See the POSIX signal API for details.