Mastering Linux Signals: How to Use kill and trap Commands Effectively

1. Linux Signal Basics

A signal is a software‑level simulation of an interrupt; sending a signal to a process triggers a handler function. A process can respond to a signal in three ways: ignore it (except SIGKILL and SIGSTOP, which cannot be ignored), catch it with a custom handler, or perform the default action defined by the kernel.

The kernel decides which response to use based on the API function used to set the handler.

Common signals (default action shown) include:

1 SIGHUP – terminate (hang up)

2 SIGINT – terminate (Ctrl+C)

3 SIGQUIT – core dump (Ctrl+\)

4 SIGILL – core dump (illegal instruction)

5 SIGTRAP – core dump (debug trap)

6 SIGABRT – core dump (abort)

7 SIGBUS – core dump (invalid address)

8 SIGFPE – core dump (floating‑point exception)

9 SIGKILL – terminate immediately (cannot be caught or ignored)

10 SIGUSR1 – terminate (user‑defined, e.g., Nginx reload)

11 SIGSEGV – core dump (invalid memory reference)

12 SIGUSR2 – terminate (user‑defined)

13 SIGPIPE – terminate (broken pipe)

14 SIGALRM – terminate (timer alarm)

15 SIGTERM – terminate (graceful shutdown, can be caught)

16 SIGSTKFLT – terminate (stack fault)

17 SIGCHLD – ignore (child process status change)

18 SIGCONT – ignore (continue stopped process)

19 SIGSTOP – stop (cannot be caught or ignored)

20 SIGTSTP – stop (Ctrl+Z)

21 SIGTTIN – stop (background read from terminal)

22 SIGTTOU – stop (background write to terminal)

23 SIGURG – ignore (out‑of‑band data)

24 SIGXCPU – core dump (CPU time limit exceeded)

25 SIGXFSZ – core dump (file size limit exceeded)

26 SIGVTALRM – terminate (virtual timer alarm)

27 SIGPROF – terminate (profiling timer)

28 SIGWINCH – ignore (window size change)

29 SIGIO – terminate (I/O ready)

30 SIGPWR – terminate (power failure)

31 SIGSYS – core dump (bad system call)

When a program crashes abnormally, the kernel writes a core dump file containing the process's memory image for debugging.

Linux defines two signal families: standard signals (1‑31, non‑realtime, not queued, may be lost) and real‑time signals (32‑64, reliable, queued).

2. Using the kill Command

The kill utility sends a specified signal to one or more processes.

kill [-s sigspec | -n signum | -sigspec] pid ...
kill -l [sigspec]

Options: -s – specify signal name -n – specify signal number -l – list signal names (1‑31 by default)

Examples:

kill -s SIGTERM 1234        # send SIGTERM by name
kill -n 15 1234             # send signal 15 (SIGTERM) by number
kill -15 1234               # shorthand
kill -SIGKILL 1234          # force termination

3. Using the trap Command

trap

defines how a shell script should react to received signals. trap [-lp] [[arg] signal_spec ...] Typical actions after catching a signal:

Remove temporary files

Ignore the signal

Prompt the user before exiting

Example 1 – ignore Ctrl+C (SIGINT) in a loop:

#!/bin/bash
trap "" 2   # 2 is SIGINT, empty arg means do nothing
for i in {1..10}; do
  echo $i
  sleep 1
done

Example 2 – exit with a message when Ctrl+C is pressed:

#!/bin/bash
trap "echo 'exit...'; exit" 2
for i in {1..10}; do
  echo $i
  sleep 1
done

Example 3 – ask the user whether to terminate the loop:

#!/bin/bash
trap "func" 2
func() {
  read -p "Terminate the process? (Y/N): " input
  if [ "$input" == "Y" ]; then
    exit
  fi
}
for i in {1..10}; do
  echo $i
  sleep 1
done

These examples illustrate how to control script termination and cleanup using signals.