Re: boot failure
On Sun, 14 Nov 2004, Kent West wrote:
> Dan Davison wrote:
> >Any help with the following would be much appreciated: I was attempting to
> >update a kernel from 2.2.20 to 2.6.8 and it seems now that neither of the
> >kernel links in the lilo menu will boot. Both exit with the same kernel
> >panic message and the output up to that point appears to be identical.
> >This output makes mention of NTFS drivers, ACPI, powerNOW and other things
> >that I believe mean I'm looking at output from the 2.6.8 kernel. I think
> >I've somehow managed to point both the links in lilo at the new kernel. Is
> >there anything that can be done, short of reinstalling debian from CD (and
> >presumably losing all data that way?)
> Yes, you can recover without reinstalling, but sometimes it's hard to
> diagnose remotely.
> Also, as an aside, if you have separate partitions for your /home (and
> possibly others), a reinstallation doesn't hurt you so bad. If it comes
> down to a reinstall and you don't have a separate partition for /home,
> you can boot off a Knoppix CD and backup your data to a network share,
> USB device, CD-ROM, etc.
> >The exit error message is (after some stuff mentioning ACPI) :
> >VFS: Cannot open root device "305" or unknown-block(3,5)
> >Please append a correct "root=" boot option
> >Kernel panic: VFS: Unable to mount root fs on unknown-block(3,5)
> >More details:
> >I obtained 2.6.8 kernel source via apt-get and included support for EXT3
> >filesystem, changed the processor type to AMD Athlon, NTFS support, SCSI
> >emulation, ACPI and processor clock scaling in menuconfig. I made a .deb
> >with make-kpkg and installed it with dpkg -i. I answered yes to the
> >question resulting from dpkg -i about making modifications to lilo so that
> >the new kernel would be available to boot. After this I checked that
> >/vmlinuz was pointing at the new kernel and looked at /etc/lilo.conf. The
> >"label" for the stanza associated with the /vmlinuz link said
> >"Linux_2.6.6" (a previous kernel installation failure)
> This makes me think the installation of your new kernel was not properly
> completed for some reason.
> > and I changed the
> >text of the label to "linux_2.6.8", but nothing else in /etc/lilo.conf.
> >After this modification to /etc/lilo.conf I ran lilo, which added the
> >various boot options OK, but produced a warning message that I did not
> >understand containing the word block and some memory addresses. Since I
> >believed that the old 2.2.20 link would be OK to boot, I proceeded to
> >power down and try to boot the new kernel, with the above-described
> >This is on an HP Athlon laptop.
> I believe what I would do is boot off a Knoppix CD (or similar) and
> manually mount the / directory rw (as necessary), redo the symlinks to
> get /vmlinuz to point to your older working 2.2 kernel, reconfigure
> /etc/lilo.conf accordingly, then run "chroot /whereverrootismounted",
> run "lilo", "exit", and "reboot". Hopefully this will give you a working
> 2.2 system, from whence you can repair what's wrong with your 2.6 kernel
> and try again.
OK thanks a lot for helpful advice (Kent and other two repliers). Knoppix
is downloaded and running. One current problem: I haven't got to run lilo
under knoppix, because it won't recognise the location of the boot
partition (specified as /dev/hda3 in lilo.conf). I have mounted and
changed write permissions on the hda3 icon on the knoppix desktop. I used
the chroot /mnt/hda5 command as suggested above by Kent. It seemed logical
to me that I'd therefore want hda3 mounted at /mnt/hda5/dev/hda3, but
I couldn't get that to happen. It seems that knoppix mounts hda3 at the
"ramdisk" by default. How do I set things up so that I can record a new
boot block according to my modified lilo.conf?
> To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact firstname.lastname@example.org