Re: loop-root (was What's Debian's /usr/src policy)
On Wed, 7 Jan 1998, Adam P. Harris wrote:
>
> [You (Dale Scheetz)]
> >On Wed, 7 Jan 1998, Turbo Fredriksson wrote:
> >> A loop-root?
>
> >With a small patch to the kernel and some modification of the loop device
> >code, you can create a file-system-in-a-file.
>
> You can do this already in stock debian (rex and hamm) with
> mount -o loop -t <fs> <file> <mount point>
>
> Why do you need to patch the kernel for this?
So that the kernel can perform this mount process for it's root file
system. It includes recognition of the kernel parameter
"loop=<loopfilesystem>" and configuration entries for enabling/disabling
the feature.
>
> >Using the loop device
> >the internal file system can be mounted as the root file system for the
> >kernel. These patches were never adopted by the kernel, and it is getting
> >harder and harder to apply them, so I currently use a patched 2.0.27
> >kernel to boot that system.
>
> Why do you have to patch the kernel at all? I already have this
> funcationality w/ initrd. Again, this is on an upatched 2.0.30 and 2.0.32
> kernels.
>
This process moves the file system into ramdisk, limiting the size of the
root file system, although there is also the possibility of mounting such
a file on that minimal ramdisk using the mount command you referenced, it
still requires some ramdisk. The loop-root doesn't.
> BTW, I use it on my custom-rolled boot floppies for debian booting off a
> floppy for a laptop which runs an initrd w/ cardmgr on it so I could get
> PCMCIA subsystem up and play with NFS-mounting root. I have a little make
> file which makes a pretty complete but small initrd file system (using
> mount -o loop) complete w/ shared libs etc, even ash. It all works, at
> least to the extent that NFS-root works well at all (which is not much but
> I haven't gotten around to pestering Joost about it).
>
This is exactly what initrd is intended to allow.
> Alternatively you could use the ramdisk and just mount root in the same
> basic fashion (although initrd is a little different -- not much really).
>
See above.
> >The one I boot with lilo is on an ext2 partition (another beauty of the
> >imbeded file system, it can be coppied to almost any other file system).
> >Interesting point: on an ext2 file system, no care need be taken. On a DOS
> >file system the file system must be un-fragmented (using dfrag) before the
> >DiD installation will work properly. The funny part of this is that it
> >doesn't matter whether you do the dfrag before you copy the file system,
> >or afterwards, it still works. Without it you get file system errors.
>
> Never noticed this issue on my 'mount -o loop' and initrd scheme. Using
> SYSLINUX to boot I think.
>
Only happens on a DOS file system, and, depending on the degree of
fragmentation of the partition, doesn't always happen there. You probably
aren't supporting such a system on a DOS partition.
Luck,
Dwarf
--
_-_-_-_-_-_- Author of "The Debian User's Guide" _-_-_-_-_-_-_-
aka Dale Scheetz Phone: 1 (904) 656-9769
Flexible Software 11000 McCrackin Road
e-mail: dwarf@polaris.net Tallahassee, FL 32308
_-_-_-_-_-_- If you don't see what you want, just ask _-_-_-_-_-_-_-
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org .
Trouble? e-mail to templin@bucknell.edu .
Reply to: