Re: install logging
Massimo Dal Zotto wrote:
> I've not explained myself. I'm not speaking of starting the install
> process from the /etc/init.d scripts, which would have the inconvenient
> that you described.
> Instead I'm proposing of using the initscript(5) optional script which,
> if present, will be used to control the start of any process spawned by
> init including the getty processes. My /etc/initscript looked like this:
>
> #
> # initscript Executed by init(8) for every program it
> # wants to spawn like this:
> #
> # /bin/sh /etc/initscript <id> <level> <action> <process>
> #
> if [ "$1" = '1' -a -x /etc/install/install.sh ]; then
> # If spawned with id '1' and /etc/install/install.sh exists run it
> exec /etc/install/install.sh < /dev/tty$1 > /dev/tty$1 2>&1
> else
> # Anything else, execute the original program
> eval exec "$4"
> fi
>
> The /etc/initscript normally doesn't exists. If the above initscript is
> copied in /etc by the bootfloppy installer after reboot it will run the
> install.sh script instead of getty, but only when handling the entry '1'
> of the first VC.
> If the script is deleted or renamed after the second stage of the
> installation is finished a getty will be started on tty1 as usual.
> While the install script is active on tty1 you still have the mgetty
> on the other VC's because the iniscript will run the original process
> if the init id is not '1'.
Huh, that's pretty slick.
If it's not too late to be making a change like this, I'd like to give
it a shot for base-config. It would be nice to not have to keep
inittab.install in synch with inittab, and should make base-config's job
less error-prone.
We'd have to make it catch the T0 and T1 id's too, for dumb
terminals.
--
see shy jo
Reply to: