"Adam J. Richter" wrote:
>
> > Useful for setting up things prior to init, like devfs
>
> If things are not set up to the point where it is safe to
> run /sbin/init, then it is not safe to run /linuxrc.
One word: RIDICULOUS
Proof of concept: LRP
> You alternative is just as "questionable", regardless of
> whether the file is named "/linuxrc" or something else.
No. It's standardized feature with properly documented 'open' behavior.
> /linuxrc and your hypothetical "preinit=" both create
> a userland context where the kernel is in a state of partial
> initialization.
Exactly. That's the point.
> The /dev/console issue does not demonstrate that /linuxrc
> is necessary, but rather that the kernel should find its initial
> console by something other than a path into a root file system,
> such as from a <major,minor> pair or a devfs name (would not require
> devfs or any file system to actually be mounted yet) modifiable
> from the boot line.
You just brought up my bitch from 2.0 series. But this will not be fixed
anytime soon.
> With this change, we could work toward
> a kernel that could be compiled without filesystem support
> to make it easier to build really tiny systems that just
> want to run some code within the kernel once everything is
> set up.
You can work for a change that may take shape over the next 3 years.
I'd like something that solves the problem in 2.4.
> So, I think renaming the /linuxrc file is not a step
> in the right direction.
I never said that. I said create a seperate kernel option, 'preinit='
that when present will be executed *similar* to /linuxrc.
/linuxrc will retain current behavior, but is considered depreciated.
By use of this option you can directly control any checks that may
work their way in, IE: preinit=/myinit,change_root:no
> If you want something to do that would be a step toward
> addressing the /dev/console issue, then perhaps you could code up
> something to enable specifying the initial console without accessing
> the filesystem.
That DOES NOT FIX THE PROBLEM. IT ONLY HELPS ONE ASPECT OF IT.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Wed May 31 2000 - 21:00:22 EST