Re: signum

[Zack Weinberg <zack@rabi.columbia.edu>]

On Mon, 2 Nov 1998 21:08:09 +0100 (CET), Mark Kettenis wrote:
>Well, defining SIGKILL to something different from 9 seems to be
>asking for trouble.  If you are going to change the numbers anyway,
>try to stick with the most common signal numbers.  It seems that all
>numbers below 15 have the same meaning among a lot of different systems.

The correspondence isn't perfect and there are a number of obsolete
signals in the 1-15 range.  These are the ones that match across
all the systems I have available:

#define SIGHUP  1       /* Hangup */
#define SIGINT  2       /* Keyboard interrupt */
#define SIGQUIT 3       /* Keyboard quit */
#define SIGILL  4       /* Illegal instruction */

#define SIGABRT 6       /* Signal raised by abort() */

#define SIGFPE  8       /* Floating point exception */
#define SIGKILL 9       /* Kill (cannot be blocked, caught, or ignored) */

#define SIGSEGV 11      /* Segmentation violation */

#define SIGPIPE 13      /* Write on a pipe with no one to read it */
#define SIGALRM 14      /* Alarm clock */
#define SIGTERM 15      /* Software termination request */

There is no correspondence above 15.  I think it would be wise to
leave a gap between the new POSIX signals and the historical
signals, and not put anything in the holes in the 1-15 range.

#define SIGSTOP 20      /* Stop (cannot be blocked, caught, or ignored) */
#define SIGCONT 21      /* Continue */
#define SIGTSTP 22      /* Keyboard stop */
#define SIGTTIN 23      /* Background read from controlling terminal */
#define SIGTTOU 24      /* Background write to controlling terminal */
#define SIGCHLD 25      /* Child terminated or stopped */

There are two historical names that should be defined, but not
if we're in a standards-compliant mode:

#define SIGIOT  SIGABRT /* IOT instruction - abort() on a PDP-11 */
#define SIGCLD  SIGCHLD /* Old System V name */

zw