Re: mysterious kernel panic at start up and it's workaround
On Wed, 06 Feb 2002 21:32:49 +1100, Paul Hampson wrote:
>On Wed, Feb 06, 2002 at 03:44:58AM -0600, Gary Turner wrote:
>> On Mon, 04 Feb 2002 16:02:23 -0800, Osamu Aoki wrote:
>>
>> >Hi,
>> >On Sun, Feb 03, 2002 at 08:50:07PM -0800, Aaron Brashears wrote:
>> >> Request_module[block-major-3] Root fs not mounted
>> >> VFS: Cannot open root device "303" or 03:03
>> >...
>> >So you can not mount root. Is root ext2?
>
>I presume it is, since the rescue disk can deal with it happily.
>
>> >> The funny thing is that if I use the append root=/dev/hda3 in
>> >> lilo.conf, it doesn't work. However if I load the debian rescue disk
>> >> and specify root=/dev/hda3, the kernel boots fine. Why won't it just
>> >> boot?
>
>That error message (to my eyes) appears to be complaining that it
>can't load the kernel module for block devices with major #3, which
>is your primary IDE bus.
>
>The problem is (I think) that the prepackaged kernels need an initrd
>to boot off, to load modules like IDE support from.
>
>To fix that, you'll have to do some reading. I compile my own kernels
>precisely because I don't want to muck about with modules or initrds. :-)
>
>> >Let's do one thing. Boot system with boot disk with root=/dev/hda3
>> >and gain root.
>
>> Now my ignorant question is--how do you pass arguments prior to or
>> during boot? As above, how does one go about booting the system off the
>> rescue disk with "root=/dev/hda3"? <--My 7th grade English teacher would
>> have had a cat :-), but doing it right looks so wrong.
>
>When you see LILO, hit shift or something. (Or turn on caps lock before
>that point)
>You should get a prompt. If you hit TAB at that point, it'll show you
>the kernel names you can boot. Type the kernel name, followed by the
>commands you want to pass it. This is the same as putting
>append=root=/dev/hda3
>in lilo.conf, which is different from putting
>root=/dev/hda3
>in lilo.conf. Subtly different, but different nonetheless.
>
>This should be mentioned on the rescue disk's opening screens. F3 I
>think...
>
>> ># vi lilo.conf
>> ># lilo
>> ># shutdown -r now
>
>> Further, it looks like you're opening vi with the file lilo.conf which
>> seems unlikely since lilo.conf is in /etc, isn't it? Is this actually
>> opening a configurator? (I don't have any man pages available at the
>> moment.) And, no editing is indicated. Then, run lilo. OK.
>> Shutdown/reboot.
>
>'vi lilo.conf' is supposed to mean 'edit /etc/lilo.conf appropriately' but
>in a rather esoteric shorthand way. :-)
I'm gonna feel pretty stupid about this, I think. Like Aaron, I could
get a rescue disk to work. No amount of fiddling with "root=" or
"append=root=" in either lilo.conf or at boot would get 2.4 up and
running. So I booted into "Linux-old" and went looking at files. I
found appropriate links in / for both the old and the new (curiously,
the kernel 2.2 link is owned by me (as regular user) while the kernel
2.4 link is owned by root). In /boot I found:
-rwxr-xr-x 1 root root ... vmlinuz-2.2.17
-rw-r--r-- 1 root root ... vmlinuz-2.4.17-686
Notice that 2.4 is not executable! I think that a chmod +x might solve
the problem. Right now, I'm too ditzy for sleep to be mucking around
with root privileges. So, goodnight.
gt
Yes I fear I am living beyond my mental means--Nash
Reply to: