[linux] vfs: cannot open root device

[Jonathan Nicol]

Heya Parker,

I'm kinda throwing things out here, but maybe this will get you on the
right track.

Do you have "/dev file system support" compiled into the kernel? I recall
it being off by default (maybe even marked experimental?) And all
appropriate IDE (or scsi) drivers?

Perhaps posting your grub.conf would help...

And, as a slightly more drastic possible solution, you could run 'make
mrproper' from /usr/src/linux (or where ever your kernel source is). This
will clean up EVERYTHING, including your config, so you'll have to start
over.


-Jonathan

On Sun, 6 Jul 2003, Parker Thompson wrote:

> I am having a problem getting a custom kernel to load and am completely
> baffled by it.
>
> I installed redhat 9 on my inspiron 2100.  I then grabbed the latest
> source (2.4.21) from kernel.org, untarred it, patched the orinoco/hermes
> drivers (for monitor mode), ran make xconfig, and ran 'make dep && make
> make && make bzImage && make modules && make modules_install'.  I then
> copied the newly created bzImage and System.map to /boot, created an
> initrd like so:
>
>   mkinitrd --fstab=/etc/fstab initrd-2.4.21 2.4.21
>
> copied it to /boot, and created an appropriate entry in grub.conf.
>
> However, when I try to boot the new kernel I get the message:
>
>   VFS: Cannot open root device "LABEL=/" or 00:00
>
> followed by the obligatory kernel panic.  The partition is ext3, for which
> I have built support into the kernel, so that is not the problem.  The
> root partition is /dev/hda1, and grub is looking at hd0,0, which should be
> fine (and mirrors my working kernel's config).
>
> Of note, some people who experienced similar problems have fixed them by
> replacing / with /dev/hda1 in grub.conf, however this did not work for me.
> I have also seen vague discussions of acpi causing problems, however
> passing the kernel acpi=off from grub had no effect in my case.
>
> Also worth mentioning, removed .config in my source root, ran make clean,
> and built a kernel with all default options (and excluding the orinoco
> patch), and encountered this same problem.  Hopefully that eliminates the
> possibility I just fouled up the configuration.
>
> Perhaps I'm just missing something silly, it's been a long time since I've
> compiled my own kernel (slacker I know), but I have run out of ideas and
> it ceased to be interesting and started to become annoying after a couple
> of hours.
>
> Thanks much for any insight,
>
> pt.
> ----------------------------------------------------------
> >From a Sun Microsystems bug report (#4102680):
> "Workaround: don't pound on the mouse like a wild monkey."
>
>
>