On 24 Mar 2001, Eric Sandeen wrote:
> On 24 Mar 2001 14:00:10 -0500, K. Mitchell Russell wrote:
> > I have tried the same thing Chris, installing the base Wolverine, and
> > it boots fine also. But still no luck with the XFS installer booting
> > either SMP or UP. I'm glad I'm not the only one, or I'd doubt my
> > sanity.
> Just talked to someone else with the same problem, and it looks like the
> installer did not create an initial ramdisk in his case - can you guys
> run "linux rescue" off the SGI boot CD and poke around to see if the
> ramdisk image exists in /boot, and whether it's called out on the
> initrd= line in /etc/lilo.conf?
> If that turns out to be the problem, instructions for re-creating the
> initrd image are at
Brilliant! I booted in 'linux rescue' mode off the SGI boot CD, created
nodes for the SCSI partitions (using block major 8), and chroot'd to my
installation. Indeed, there are NO initrd's on my system whatsoever, and
no entries for them in /etc/lilo.conf. I created the initrd's and
booted perfectly into my new installation.
> > Thanks for the info Eric. I don't know what kind of magic words you
> > used at boottime, but I'm still reproducing the problem on my
> > hardware.
> I just realized that although I did the install on SCSI, the actual boot
> device is still /dev/hda1 (i.e. lilo is installed on /dev/hda1) so
> that's kind of cheating. I'll turn off the IDE controllers and re-try
> > Are you using any bootprompt commands? I'll keep trying,
> > but since I can't even get the system up after running the XFS
> > installer, I can't very well look at the results of the installer to
> > determine the problem.
> You can try running "linux rescue" off the CD to poke around your system
> a bit...
Never had the occasion to use it until now. Learn something new every
> > I do know that the kernel panic is at the
> > point where the VFS is loading root from the initrd, after the ramdisk
> > is loaded, and before the scsi modules are loaded, not the point of
> > where it's loading the XFS scsi disk root.
> Hm, then perhaps the initrd image _is_ getting created... maybe you can
> verify this with "linux rescue"
Indeed they are not being created. I'm a relative newbie, and not
familiar enough with the anaconda to see why, but I did notice at the
end of the installation, it reported problems unmounting sysimage,
sysimage/usr, and tmp/loop0 (failed: 16). Is loop0 being used to create
the initrd perhaps?
Thanks for the help,
K. Mitchell Russell, M.D.
MedITAC Research Lab