Re: runit on debian sid success, and questions

[Thomas Schwinge <schwinge-lists-skarnet.org-supervision@nic-nac-project.de>]

On Tue, Jan 18, 2005 at 10:07:44PM -0800, Larry Doolittle wrote:
> On Tue, Jan 18, 2005 at 05:13:03PM -0800, Larry Doolittle wrote:
> > 
> > I ran into odd runit shutdown behavior on another machine,
> > vanilla Pentium (Celeron) with Debian sarge.  I don't have
> > the exact message, but it said something like
> > 
> >    power down
> >    kernel panic, attempting to kill init!
> > 
> > then the disks spun down, but the video kept going.
> > It then needed me to reset or power down.  Of course, this
> > machine powered down normally with stock debian sarge init.
> 
> OK, I lied.  It powered down normally with Red Hat 7.3, but there
> is a similar lack-of-power-down with stock debian sarge init.
> Still, there is an extra wrinkle with ruinit.
> 
> Transcript (by hand, I tried to avoid any errors):
> 
> ...normal shutdown stuff...
> - runit: power off
> flushing ide devices: hda hdc
> Power down.
> Kernel panic: Attempted to kill init!
> 
> <whine as disk spins down>
> <caps log & scroll lock LEDs start blinking with one second period>
> <power stays on, screen is still there, and reacts to ALT-Fn>

#v+
[...]
From: Gerrit Pape <pape@smarden.org>
To: supervision@list.skarnet.org
Subject: Re: Shutdown kernel panic
[...]

On Mon, Feb 02, 2004 at 09:27:17PM -0800, Anthony Gorecki wrote:
> Using RunIt on kernel 2.6.1, I receive the following output at the end of 
> executing init 0:
> 
> - - runit: leave stage: /etc/runit/3
> - - runit: sending KILL signal to all processes...
> - - runit: power off...
> Power down.
> Kernel panic: Attempted to kill init!

I've seen this too with older Linux kernels.  I guess it depends on the
kernel configuration.  It shouldn't hurt much, at this time everything
is ready for power off.
[...]
#v-

#v+
[...]
From: yoppo@jippii.fi
To: supervision@list.skarnet.org
Subject: about runit...
[...]

Hi!

First I must say that I started using runit about month ago and I must=20
say I am pretty pleased with it. Now I think I found small bug in it.=20
When acpi is disabled (acpi=3Doff at lilo append) I got kernel panic at=20
end of the shutdown/reboot process:

....
-runit: power off...
Power down.
Kernel panic: Attempted to kill init!

Off course this actually doesn't matter much but I think it should cause=20
this when acpi disabled.
[...]
#v-

> The only extra stuff here is the "Kernel panic: Attempted to kill init!"
> message.  Does sysv.init spin in an idle loop if the power down fails?

At a _quick_ glance it doesn't seem to.
sysvinit-2.85, halt.c
#v+
[...]
        if (do_hddown)
                (void)hddown();

        if (do_reboot) {
                init_reboot(BMAGIC_REBOOT);
        } else {
                /*
                 *      Turn on hard reboot, CTRL-ALT-DEL will reboot now
                 */
                init_reboot(BMAGIC_HARD);

                /*
                 *      Stop init; it is insensitive to the signals sent
                 *      by the kernel.
                 */
                kill(1, SIGTSTP);

                /*
                 *      Halt or poweroff.
                 */
                if (do_poweroff)
                        init_reboot(BMAGIC_POWEROFF);
                /*
                 *      Fallthrough if failed.
                 */
                init_reboot(BMAGIC_HALT);
        }

        /*
         *      If we return, we (c)ontinued from the kernel monitor.
         */
        init_reboot(BMAGIC_SOFT);
        kill(1, SIGCONT);

        exit(0);
}
#v-


Regards,
 Thomas