tisdagen den 25 september 2001 10.15 skrev du:
> Oden Eriksson <[EMAIL PROTECTED]> writes:
>
>
> [...]
>
> > > > on MDK7.2. This machine works fine without a "/initrd" directory. (it
> > > > uses a initrd for the reiserfs module and a loop module).
> > >
> > > mkinitrd is different we need that to have ext3/xfs/jfs / filesystem
> > > (and we do a pivot_root) i'm sure gc can explain more about that.
> >
> > Ok thanks. I read the kernel docs about it and it seems complicated... I
> > will examine the latest mkinitrd script though.
> >
> > Anyhow... If this directory is so crucial for MDK8.1, its existence
> > should be verified somewhere, maybe in the "halt" script, in lilo, in the
> > "installkernel" script, linuxconf etc. Or is this done allready that I
> > don't know of?
>
> No, it's not :-(.
>
> For post-8.1, my idea was to use a /lib/initrd directory so that people
> will not notice it and remove it by error.
Yes, a good idea.
> I don't know if it would be really nice to verify it in `halt', in lilo
> it's not possible, in the initrd I can't create it since we're mounted RO,
> and in installkernel it's not relevant since we seldomly run this script.
Why not verify and create if needed the "/initrd" dir from the "mkinitrd"
script for now? That would be a quick fix, and maybe even notify the user NOT
to delete the "/initrd" dir ;)
I miss "chattr" for RFS...
--
Regards // Oden Eriksson
