grub errors and read-only

Donal Farrell vmlinuz101 at gmail.com
Fri Aug 5 13:21:17 PDT 2005


I put the following from suse's /boot into FC3's /boot 
LFS vmlinuz, system.map and config
SuSE vmlinuz, system.map and config

I also changed suse's menu.lst to have root=/dev/hda1 

i booted into suse and got VFS kernel panic

I changed it back to root=/dev/hda2 and got the follwoing

ext2_fs warning -> device hda2
ext2_fill_super -> mounting ext3 on ext2
VFS mounted root (ext2 fs) readonly
warning unable to open an initial console
kernel panic attempted to kill init




On 8/5/05, Ken Moffat <ken at kenmoffat.uklinux.net> wrote:
> On Fri, 5 Aug 2005, Donal Farrell wrote:
> 
> > yupe, i tried to put sarge on /dev/hda2 which didn't exist at all when
> > i just had suse and lfs. the /dev/hda2 was the primary partition i
> > made for debian, and thought you can't mount this on / as suse is on /
> > on /dev/hda1 (am I totally wrong here?).
> 
>  You probably can't mount over '/' because your shell is running under
> '/' (loose terminology), so I think I agree.
> 
> > therefore i chose to make a
> > /boot on /dev/hda2 which i take to mean that sarge will be mounted on
> > /boot on /dev/hda2. the debian installer kicks out but only after it
> > partitions the drive.
> 
>  If the host system mounts something at /boot (which is a standard
> directory), and that something is another install, then
> 
> (1.) the existing contents of /boot are inaccessible
> (2.) the files from the (debian) install are at /boot /boot/boot
> /boot/bin /boot/etc /boot/usr and so forth, which is usually not very
> convenient.
> 
> > As per instructions in lfs, the lfs grub
> > bootloader takes precedence over all otheres, by which i mean it loads
> > itself or suse or whatever. this menu.lst is
> >
> > title LFS 6.0
> > root (hd0,0)
> > kernel /boot/lfskernel-2.6.8.1 root=/dev/hda1
> >
> > why is root on /dev/hda1 when lfs is on /dev/hda4??
> 
>  Like I said, I'm *not* a grub user (well, I've still got it on one box
> at the moment) but I imagine that is telling grub to use hda1 (0,0) to
> find that kernel, and then telling the kernel that hda1 is '/' which
> sounds very wrong - if you aren't using modules, that might work
> (running the LFS kernel on Suse).
> 
> > also, from the same menu.lst, suse is
> >
> > title SuSE Linux 9.1 Professional
> >     kernel (hd0,1)/boot/vmlinuz root=/dev/hda2 vga=0x31a splash=silent
> > desktop resume=/dev/hda5 showopts
> >     initrd (hd0,1)/boot/initrd
> >
> > this shows a swap on /dev/hda5, which i deleted and put on /dev/hda3
> > and shows root=/dev/hda2 :(
> >
> 
>  Wow, now that's going over my head : 0,2 is probably hda2 ?  Sounds as
> if Suse thinks it's on hda2.  If Suse and fedora both use initrds, you
> will need to rename at least one of them in case you end up putting
> them in the same /boot/ directory (also edit the corresponding grub
> entry).  BUT, mixing software suspend with recovering from disaster is
> asking for pain - I don't recognise all of those options, but resume=
> will cause pain if wherever you point it to doesn't contain the stored
> system image (which it won't after booting a different system and using
> swap).
> 
> > the actual menu.lst in suse's grub is
> >
> > title SuSE Linux Professional 9.1
> >     kernel (hd0,0)/boot/vmlinuz root=/dev/hda1 vga=0x31a splash=silent
> > desktop resume=/dev/hda3 showopts
> >     initrd (hd0,0)/boot/initrd
> >
> > here i can see the valid swap
> >
> > by the way, i only have suse, FC4 and lfs, along with the swap now. By
> > right, I should have shared /home and by right i should have extended
> > partitions.
> >
> > do i HAVE to overwrite suse now and reinstall it or is there indeed
> > some magic i can perform here? what must i do in /etc/fstab in fedora
> > to allow entries for suse and lfs? by the way, i also have xen, bochs
> > and vmware on suse and their kernel images are also in /boot
> >
> 
>  It all depends.  You shouldn't need to overwrite Suse, just change
> various things within it.  But, to be able to repartition, you need
> unallocated space on the drive.  Copying to another drive, or perhaps
> writing to CD or DVD, could preserve the data, but you then need a
> bootable CD that you are comfortable with to repartition and restore the
> data [ using a rescue CD plus a CD/DVD of backed-up data in the *same*
> drive imposes severe constraints on the rescue CD and forces it to use a
> very minimal busybox and uclibc, or a minimal shell with dietlibc ] so
> you will probably need to reinstall after making and testing backups.
> 
>  Then decide on the required partition layout [ perhaps with multiple
> dedicated swap partitions if you are going to try suspend-to-disk,
> perhaps with a partition to eventually share as /boot ], then
> reinstalling one distro and tweaking its partitioning.  This is NOT
> something to rush into.
> 
>  You've reached the limits of my understanding about how grub works, but
> I still believe you can probably boot the installed Suse and LFS using
> the type of procedure I outlined earlier.  BUT, you need to become clear
> where grub is installed, which partition is which, and how grub sees
> them.  IF your last install was fedora AND you let it set up grub, THEN
> go with what is in fedora's /boot (don't rename any of the fedora
> kernels or initrds) and try to add the others.  After each attempt to
> add them, reboot, see if the extra system boots, and if it does, see if
> it's using the expected system (fedora probably has /etc/redhat-release
> or similar, Suse will have something unique, LFS should have
> /etc/lfs-release).  You'll still need to fix the fstabs in the Suse and
> LFS systems before trying to boot them.
> 
> Ken
> --
>  das eine Mal als Tragödie, das andere Mal als Farce
> 
> --
> http://linuxfromscratch.org/mailman/listinfo/blfs-support
> FAQ: http://www.linuxfromscratch.org/blfs/faq.html
> Unsubscribe: See the above information page
>



More information about the blfs-support mailing list