SIGNAL(9)

πŸ“¦ version 15.1, in freebsd 14.5, freebsd 15.1

πŸ’– SIGNAL(9) β€” Kernel Developer's Manual πŸ’–

okay so bestie, welcome to SIGNAL(9), the kernel signal functions page from the Kernel Developer’s Manual. This is the whole signal-wrangling squad: macros for poking at signal sets, functions for delivering signals, and all the locking lore that comes with them. no cap, it’s iconic kernel plumbing. πŸ’…


✨ NAME ✨

signal, SIGADDSET, SIGDELSET, SETEMPTYSET, SIGFILLSET, SIGISMEMBER, SIGISEMPTY, SIGNOTEMPTY, SIGSETEQ, SIGSETNEQ, SIGSETOR, SIGSETAND, SIGSETNAND, SIGSETCANTMASK, SIG_STOPSIGMASK, SIG_CONTSIGMASK, SIGPENDING, cursig, execsigs, issignal, killproc, pgsigio, postsig, sigexit, siginit, signotify, trapsignal β€” kernel signal functions.


🌸 SYNOPSIS 🌸

#include <sys/param.h>
#include <sys/proc.h>
#include <sys/signalvar.h>

void
SIGADDSET(sigset_t set, int signo);

void
SIGDELSET(sigset_t set, int signo);

void
SIGEMPTYSET(sigset_t set);

void
SIGFILLSET(sigset_t set);

int
SIGISMEMBER(sigset_t set, int signo);

int
SIGISEMPTY(sigset_t set);

int
SIGNOTEMPTY(sigset_t set);

int
SIGSETEQ(sigset_t set1, sigset_t set2);

int
SIGSETNEQ(sigset_t set1, sigset_t set2);

void
SIGSETOR(sigset_t set1, sigset_t set2);

void
SIGSETAND(sigset_t set1, sigset_t set2);

void
SIGSETNAND(sigset_t set1, sigset_t set2);

void
SIG_CANTMASK(sigset_t set);

void
SIG_STOPSIGMASK(sigset_t set);

void
SIG_CONTSIGMASK(sigset_t set);

int
SIGPENDING(struct proc *p);

int
cursig(struct thread *td);

void
execsigs(struct proc *p);

int
issignal(struct thread *td);

void
killproc(struct proc *p, char *why);

void
pgsigio(struct sigio **sigiop, int sig, int checkctty);

void
postsig(int sig);

void
sigexit(struct thread *td, int signum);

void
siginit(struct proc *p);

void
signotify(struct thread *td);

void
trapsignal(struct thread *td, int sig, u_long code);

πŸŽ€ DESCRIPTION πŸŽ€

  • SIGADDSET() β€” adds signo to set. And lol, no effort is made to ensure that signo is a valid signal number. it’s giving β€œtrust me bro”.
  • SIGDELSET() β€” removes signo from set. Same energy: no effort is made to ensure that signo is a valid signal number.
  • SIGEMPTYSET() β€” clears all signals in set. Empty set, clean slate, main character reset ✨.
  • SIGFILLSET() β€” sets all signals in set. Full set, no crumbs.
  • SIGISMEMBER() β€” determines if signo is set in set.
  • SIGISEMPTY() β€” determines if set does not have any signals set.
  • SIGNOTEMPTY() β€” determines if set has any signals set.
  • SIGSETEQ() β€” determines if two signal sets are equal; that is, the same signals are set in both.
  • SIGSETNEQ() β€” determines if two signal sets differ; that is, if any signal set in one is not set in the other.
  • SIGSETOR() β€” ORs the signals set in set2 into set1.
  • SIGSETAND() β€” ANDs the signals set in set2 into set1.
  • SIGSETNAND() β€” NANDs the signals set in set2 into set1.
  • SIG_CANTMASK() β€” clears the SIGKILL and SIGSTOP signals from set. These two signals cannot be blocked or caught, and SIG_CANTMASK() is used in code where signals are manipulated to ensure this policy is enforced. this is a hard boundary, bestie.
  • SIG_STOPSIGMASK() β€” clears the SIGSTOP, SIGTSTP, SIGTTIN, and SIGTTOU signals from set. SIG_STOPSIGMASK() is used to clear stop signals when a process is waiting for a child to exit or exec, and when a process is continuing after having been suspended.
  • SIG_CONTSIGMASK() β€” clears the SIGCONT signal from set. SIG_CONTSIGMASK() is called when a process is stopped.
  • SIGPENDING() β€” determines if the given process has any pending signals that are not masked. If the process has a pending signal and the process is currently being traced, SIGPENDING() will return true even if the signal is masked. sneaky but documented.
  • cursig() β€” returns the signal number that should be delivered to process td->td_proc. If there are no signals pending, zero is returned.
  • execsigs() β€” resets the signal set and signal stack of a process in preparation for execve(2). The process lock for p must be held before execsigs() is called.
  • issignal() β€” determines if there are any pending signals for process td->td_proc that should be caught, or cause this process to terminate or interrupt its current system call. If process td->td_proc is currently being traced, ignored signals will be handled and the process is always stopped. Stop signals are handled and cleared right away by issignal() unless the process is a member of an orphaned process group and the stop signal originated from a TTY. The process spin lock for td->td_proc may be acquired and released. The sigacts structure td->td_proc->p_sigacts must be locked before calling issignal() and may be released and reacquired during the call. The process lock for td->td_proc must be acquired before calling issignal() and may be released and reacquired during the call. Default signal actions are not taken for system processes and init. locking lore unlocked πŸ”.
  • killproc() β€” delivers SIGKILL to p. why is logged as the reason why the process was killed. bye bestie, and it’s on the record.
  • pgsigio() β€” sends the signal sig to the process or process group sigiop->sio_pgid. If checkctty is non-zero, the signal is only delivered to processes in the process group that have a controlling terminal. If sigiop->sio_pgid is for a process (> 0), the lock for sigiop->sio_proc is acquired and released. If sigiop->sio_pgid is for a process group (< 0), the process group lock for sigiop->sio_pgrp is acquired and released. The lock sigio_lock is acquired and released.
  • postsig() β€” handles the actual delivery of the signal sig. postsig() is called from ast() after the kernel has been notified that a signal should be delivered (via a call to signotify(), which causes the flag PS_NEEDSIGCHK to be set). The process lock for process that owns curthread must be held before postsig() is called, and the current process cannot be 0. The lock for the p_sigacts field of the current process must be held before postsig() is called, and may be released and reacquired.
  • sigexit() β€” causes the process that owns td to exit with a return value of signal number sig. If required, the process will dump core. The process lock for the process that owns td must be held before sigexit() is called.
  • siginit() β€” is called during system initialization to cause every signal with a default property of SA_IGNORE (except SIGCONT) to be ignored by p. The process lock for p is acquired and released, as is the lock for sigacts structure p->p_sigacts. The only process that siginit() is ever called for is proc0. special little boot-time behavior.
  • signotify() β€” flags that there are unmasked signals pending that ast() should handle. The process lock for process td->td_proc must be held before signotify() is called, and the thread lock is acquired and released.
  • trapsignal() β€” sends a signal that is the result of a trap to process td->td_proc. If the process is not being traced and the signal can be delivered immediately, trapsignal() will deliver it directly; otherwise, trapsignal() will call psignal(9) to cause the signal to be delivered. The process lock for td->td_proc is acquired and released. The lock for the p_sigacts field of td->td_proc is acquired and released.

🌈 RETURN VALUES 🌈

The SIGISMEMBER(), SIGISEMPTY(), SIGNOTEMPTY(), SIGSETEQ(), SIGSETNEQ(), and SIGPENDING() macros all return non-zero (true) if the condition they are checking is found to be true; otherwise, zero (false) is returned.

The cursig() function returns either a valid signal number or zero.

issignal() returns either a valid signal number or zero.


πŸ¦‹ SEE ALSO πŸ¦‹

pgsignal(9), psignal(9) β€” go read the friends, iykyk πŸ“š.


πŸ‘‘ AUTHORS πŸ‘‘

This manual page was written by Chad David <davidc@FreeBSD.org>.


xoxo bestie πŸ’– β€” Debian β€” July 14, 2023 β€” SIGNAL(9) πŸŽ€

πŸ“¦ where this page comes from