[Pkg-uml-pkgs] [rootstrap] how to proceed?

[Stefano Melchior]

On Mon, May 15, 2006 at 10:50:45PM +0200, Mattia Dongili wrote:
Dear all,
> > I tried to investigate how rootstrap works and, despite of the fact I did
> > know much about python language, I realized the principles of its idea.
> > 
> > At this moment it fails, above all, because the current method does not
> > support anymore or in the proper way the definition on /dev/ubdn devices
> > well on the host you attempt to mount/unmount the uml filesystem.
> > It launches a linux (indeed the uml kernel) instance to configure the
> > newly created filesystem.
> 
> The problem is basically that rootstrap mounts host's / as its own / and
> /dev is missing proper device nodes (see link below).
> Using udev won't probably solve /dev/ubd problems as it might conflict
> with an existing udev running on the host (? not tried actually).
> 
> > The rest seems to work fine.
> > I was wondering if:
> > - it is reasonable to patch the current rootstrap to make it works on
> >   /dev/ubdn device;
> 
> If it's needed to make rootstrap work then it is reasonable.

there is a patch regarding this, which it was applied, but rootstrap has still 
been affected by the problem of /dev/ubd
> 
> > - it is reasonable to make the package not to be dependant on
> >   user-mode-linux and to make it create a filesystem by using only 
> >   deboostrap, then you can automate the process of set the proper config
> >   files up.
> 
> This is exactely what I was suggesting here:
> http://lists.alioth.debian.org/pipermail/pkg-uml-devel/2006-January/000126.html
> I think it's way easier generating the rootfs in a loop mounted image.
> From the manpage though, I read the the big advantage in current
> approach is that it's not necessary to run it as root. Hmmm... this
> would be a big loss.

yes, it really is, that is the great advantage of using uml linux command
versus the other possible way where you are asked to run a command by
root.
> 
> I'm wondering if it makes sense to generate a minimal initrd first using
> host binaries and then let rootstrap use this to load the initial fs and
> boostrap filesystem creation.
> But probably once you have to setup an initrd you'd better setup the
> full system...
> 
> BTW: I'll probably try to rewrite rootstrap in Perl one day, I hate not
> being able to hack on it :)
it may be the last pkg, but it is a good idea to fix its main issue, I
mean to make it work. Doesn't it matter how long it takes.

I shall try to work on in my few spare time.

SteX
-- 
Stefano Melchior, GPG key = D52DF829 - <stefano.melchior at openlabs.it>
http://etinarcadiaego.dyndns.org    --     http://www.stex.name
Skype ID "stefanomelchior"
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 307 bytes
Desc: Digital signature
Url : http://lists.alioth.debian.org/pipermail/pkg-uml-pkgs/attachments/20060516/6ba497a2/attachment-0001.pgp