[rock-devel] On system init scripts
Stefan Fiedler
stefan.fiedler at students.jku.at
Thu Sep 7 22:53:09 CEST 2006
> On Patch [2006082512411919808]:
>
> Stefan Fiedler:
> split system.init and network.init scripts into several smaller ones
>
> new msg by stefanp:
> - the new /etc/initscript does not seem to do anything of effect anymore,
> in which case it should be deleted.
It sets umask and ulimit and redirects standard error to standard output for
all programs run by init.
>
> - devfs.init should probably be merged with devfsd.init package.
devfsd.init starts an optional user-space program. devfs.init enables a kernel
feature, so it would better fit to the linux24 package.
>
> - the new execle() call in rc.c is wrong.
Indeed. /bin/sh generally does not support the -l option...
>
> - static-dev.init must be a practical joke? the whole point of a static
> /dev is that it's always there on disk and does not need anything to be
> done, least of all having an empty ramfs mounted over it!
The ramfs mount is a fallback when there is no console device file on the
(read-only) root partition.
That code was already in there and I didn't dare to remove it.
>
> - the changes to sysfiles/etc_rc.d_rc.txt seem arbitrary to me, but oh
> well ...
init sets PATH=/bin:/usr/bin:/sbin:/usr/sbin (for all it's children), while rc
sets the path to /bin:/sbin:/usr/bin:/usr/sbin. Having a different path
search order for children of init and children of rc is probably no good, so
rc does not redefine PATH anymore.
>
> - why use an /etc/rc.d/sysinit that does not do anything?
Just to show we can. :)
>
> - what is the motivation for the change in 00-dirtree/data.txt? it will be
> gone again when a /dev gets mounted.
This is for the static /dev directory. Without /dev/initctl one cannot change
runlevels with the init program
>
> - the changes to devfsd/devfsd.conf must be a mistake?
That was a simple test case for my fl_wrapper patch. Will be removed before
the patch is applied.
Thanks for your feedback. There will probably be more changes to this patch
before it is applied.
Greetings,
Stefan Fiedler
> You can't separate peace from freedom because no one can be at peace unless
> he has his freedom. Malcolm X, Malcolm X Speaks, 1965
-------------- next part --------------
SubMaster Action:
new msg by stefanp:
- the new /etc/initscript does not seem to do anything of effect anymore, in which case it should be deleted.
- devfs.init should probably be merged with devfsd.init package.
- the new execle() call in rc.c is wrong.
- static-dev.init must be a practical joke? the whole point of a static /dev is that it's always there on disk and does not need anything to be done, least of all having an empty ramfs mounted over it!
- the changes to sysfiles/etc_rc.d_rc.txt seem arbitrary to me, but oh well ...
- why use an /etc/rc.d/sysinit that does not do anything?
- what is the motivation for the change in 00-dirtree/data.txt? it will be gone again when a /dev gets mounted.
- the changes to devfsd/devfsd.conf must be a mistake?
On Patch [2006082512411919808]:
Stefan Fiedler:
split system.init and network.init scripts into several smaller ones
--
https://www.rocklinux.net/submaster/?mode=patch&uid=2006082512411919808
More information about the rock-devel
mailing list