correct the init deps article

Signed-off-by: Nico Schottelius <nico@ikn.schottelius.org>
This commit is contained in:
Nico Schottelius 2009-08-05 21:12:28 +02:00
parent 96ee04efe4
commit 984f1f5846

View file

@ -1,14 +1,13 @@
[[!meta title="About init dependencies"]] [[!meta title="About init dependencies"]]
As i started to hack on [cinit](http://unix.schottelius.org) again, As i started to hack on [cinit](http://unix.schottelius.org/cinit/) again,
I tried to get it running on Debian in a VM. I tried to get it running on Debian in a VM.
I took the old configuration from my last computer and tried to boot I took the old configuration from my last computer and tried to boot
with cinit, which failed, because the udev stuff changed. with cinit, which failed, because the udev stuff changed.
So I added a udev service, which uses /etc/init.d/udev for switching So I added a udev service, which uses /etc/init.d/udev for switching on.
on.
After booting the VM, I recognized, that the service After booting the VM, I recognized, that the service
***mount/proc*** (which mounts /proc...) fails. Mount claims that ***mount/proc*** (which mounts /proc...) fails. Mount claims that
@ -22,13 +21,13 @@ Besides the problematic of using a sys-v-init script with an
intelligent init system, it's interesting to see, **why** intelligent init system, it's interesting to see, **why**
this happens: this happens:
The ***service*** mount/root **needs** the device files in /dev The service mount/root needs the device files in /dev
to be able to run mount/root/fsck (the filesystem check). to be able to run mount/root/fsck (the filesystem check).
Thus mount/root requires the service ***udev*** to be started before. Thus mount/root requires the service udev to be started before.
The ***service*** mount/proc **needs** the root filesystems writable, The service mount/proc needs the root filesystems writable,
to write into /etc/mtab and thus **needs** mount/root. to write into /etc/mtab and thus needs mount/root.
This also demonstrates the problematic of the historical grown This also demonstrates the problematic of the historical grown
init scripts, which do way more than one job and their lack of init scripts, which do way more than one job and their lack of