Never returning inttab line

[Denys Vlasenko]

On Saturday 31 May 2014 01:42, Felipe de Andrade Neves Lavratti wrote:
> Recently I started working in a system where there is a never returning
> call in the busybox inttab file, the line is under the ::sysinit directive.
> 
> What are the consequences of this? Respawns are going to work properly?

init.c has a comment explaining it in detail:

/* Each type of actions can appear many times. They will be
 * handled in order. RESTART is an exception, only 1st is used.
 */
/* Start these actions first and wait for completion */
#define SYSINIT     0x01
/* Start these after SYSINIT and wait for completion */
#define WAIT        0x02
/* Start these after WAIT and *dont* wait for completion */
#define ONCE        0x04
/*
 * NB: while SYSINIT/WAIT/ONCE are being processed,
 * SIGHUP ("reread /etc/inittab") will be processed only after
 * each group of actions. If new inittab adds, say, a SYSINIT action,
 * it will not be run, since init is already "past SYSINIT stage".
 */
/* Start these after ONCE are started, restart on exit */
#define RESPAWN     0x08
/* Like RESPAWN, but wait for <Enter> to be pressed on tty */
#define ASKFIRST    0x10
/*
 * Start these on SIGINT, and wait for completion.
 * Then go back to respawning RESPAWN and ASKFIRST actions.
 * NB: kernel sends SIGINT to us if Ctrl-Alt-Del was pressed.
 */
#define CTRLALTDEL  0x20
/*
 * Start these before killing all processes in preparation for
 * running RESTART actions or doing low-level halt/reboot/poweroff
 * (initiated by SIGUSR1/SIGTERM/SIGUSR2).
 * Wait for completion before proceeding.
 */
#define SHUTDOWN    0x40
/*
 * exec() on SIGQUIT. SHUTDOWN actions are started and waited for,
 * then all processes are killed, then init exec's 1st RESTART action,
 * replacing itself by it. If no RESTART action specified,
 * SIGQUIT has no effect.
 */
#define RESTART     0x80


In short: no. Until sysinit finishes, RESPAWN actions will not be started.